Bioelectronics for electrical stimulation: materials, devices and biomedical applications

Abstract

Bioelectronics is a hot research topic, yet an important tool, as it facilitates the creation of advanced medical devices that interact with biological systems to effectively diagnose, monitor and treat a broad spectrum of health conditions. Electrical stimulation (ES) is a pivotal technique in bioelectronics, offering a precise, non-pharmacological means to modulate and control biological processes across molecular, cellular, tissue, and organ levels. This method holds the potential to restore or enhance physiological functions compromised by diseases or injuries by integrating sophisticated electrical signals, device interfaces, and designs tailored to specific biological mechanisms. This review explains the mechanisms by which ES influences cellular behaviors, introduces the essential stimulation principles, discusses the performance requirements for optimal ES systems, and highlights the representative applications. From this review, we can realize the potential of ES based bioelectronics in therapy, regenerative medicine and rehabilitation engineering technologies, ranging from tissue engineering to neurological technologies, and the modulation of cardiovascular and cognitive functions. This review underscores the versatility of ES in various biomedical contexts and emphasizes the need to adapt to complex biological and clinical landscapes it addresses.

---

Ya Huang

Dr Ya Huang is currently a Senior Research Fellow at the Department of Biomedical Engineering of City University of Hong Kong (CityU). She received her BS degree (2015) and PhD degree (2020) in Materials Science and Engineering from Tsinghua University. She is trained as a scientist with interdisciplinary backgrounds, including materials science, physics, electronics and biomedical engineering. Her current research interests mainly focus on wearable haptic feedback systems and miniaturized electrostimulation devices. She has published over 40 papers in prestigious journals like Nature, Nature Electronics, Science Advances, etc.

Kuanming Yao

Kuanming Yao is currently a postdoctoral researcher in the Department of Electrical & Computer Engineering in University of California, Los Angeles (UCLA). He was a postdoc in Prof. Xinge Yu's lab in the Department of Biomedical Engineering, City University of Hong Kong from 2023 to 2024, where he also obtained his PhD degree in 2023. His research interests include manufacturing techniques of soft electronics, integrated bio-electronics for in vivo metabolic monitoring and therapeutic stimulations, and wearable haptic interfaces. He has cumulatively published over 50 journal articles, and is serving as first/co-first author in more than 10 papers in journals including Nature Machine Intelligence, Nature Electronics, Science Advances, etc.

Qiang Zhang

Dr Qiang Zhang is a postdoctoral fellow at the Department of Biomedical Engineering, CityU. He received his PhD in Biomedical Engineering from PolyU (2022) and Bachelor of Clinical Medicine (2016) and Master of Public Health (2018) from XJTU. He is trained as a scientist with interdisciplinary backgrounds, including medicine, materials science, chemistry and electronics. His research focuses on development of novel bioelectronic devices that integrate technologies of tissue engineering and flexible electronics to promote functional tissue repair while achieving real-time disease monitoring. He has published over 40 papers in prestigious journals like Nature, Nature Photonics, Nature Communications, Science Advances, and ACS Nano.

Xingcan Huang

Xingcan Huang is currently a PhD student at the Department of Biomedical Engineering, City University of Hong Kong. He received his MPhil degree and bachelor's degree from Shandong University and Qingdao University in 2020 and 2017, respectively. His current research interests include flexible and stretchable electronics, flexible biosensors and flexible nanomaterials and their applications in intelligent biomedical engineering.

Zhenlin Chen

Zhenlin Chen is currently a PhD candidate in Professor Xinge Yu's group at the Department of Biomedical Engineering, City University of Hong Kong. He received his BE degree from Harbin Engineering University in 2017 and his ME degree from the Harbin Institute of Technology in 2020. He has an interdisciplinary background in mechanics, electronics, fluid mechanics and electromagnetism. His research focuses on microfabrication, human–robot interaction, and microfluidics for wearable electronics.

Xinge Yu

Xinge Yu is the Associate Director of Institute of Digital Medicine of City University of Hong Kong, a Member of the Hong Kong Young Academy of Sciences, the recipient of RGC Research Fellow, Innovators under 35 China (MIT Technology Review) and Stanford's top 2% most highly cited scientists, etc. Xinge Yu's research group is focusing on skin-integrated electronics and systems for VR and biomedical applications. Dr Yu serves as an Associate Editor and a member of Editor Boards for 12 journals. He has published 180 papers in Nature, Nature Materials, Nature Electronics, etc. and has filed/granted 50 patents.

---

1. Introduction

Bioelectronics integrates various concepts ranging from materials science, physics, chemistry, and biology to medicine, focusing on the development and exploration of new electronic devices and technologies to interact profoundly with biological systems.^1–3 One of the most important techniques in bioelectronics is electrical stimulation (ES), a versatile and powerful technique used across a broad spectrum of medical and therapeutic applications.⁴ By delivering controlled electrical impulses to specific areas of the body, ES can modulate cellular and tissue functions, offering significant benefits in both clinical and research settings.⁵ ES provides a potent means to regulate cellular behaviors and tissue functionalities, aiming to rectify or enhance physiological functions disrupted by diseases or injuries.^6–11 By delivering electrical currents or fields, ES based bioelectronics can activate or suppress biological activities at the molecular, cellular, tissue, and organ levels.^12–16 All living organisms possess an intricate network of bioelectrical signals, where neurons fire, hearts beat, and muscles contract, all governed by ion flows and electrical impulses.^4,17–20 ES has been widely adopted for its ability to influence neural activity, facilitate muscle contraction, and promote tissue repair and regeneration.¹⁸ This technique is particularly valuable in fields such as neurology, where it is used to treat conditions like chronic pain, epilepsy, and Parkinson's disease, as well as in rehabilitation medicine, where it aids in muscle re-education and recovery from injuries.^21,22 At its essence, the field of bioelectronics for ES aims to manipulate the inherent electrical properties of biological tissues for diagnostic, therapeutic, and reparative purposes.^5,12,23

Developing bioelectronics for ES hinges on the strategic integration of electrical signals, device interfaces and design to achieve optimal performances.^11,24,25 The key to this process is the selection of appropriate electrical signals, which necessitates a deep understanding of the biological mechanisms in target applications.^4,8 This foundation allows for the precise tuning of parameters or the adoption of new ES principles. Additionally, the bioelectronics must be capable of generating robust electrical outputs, characterized by sufficient voltage and lifespan, tailored to specific uses. A critical aspect of effective ES bioelectronic systems is the interface between the electronics and the biological tissues.^26–28 This interface ensures efficient transmission of electrical pulses, which is vital for the intended biological responses. The design considerations for bioelectronics extend to the integration of device architecture and the optimization of its size, structure, performance, and biocompatibility, ensuring that the device functions harmoniously within a biological environment.^29–31 Unconventional device design has transformed bioelectronics into soft and ultrathin forms that conform to curved biological surfaces, enabling high-quality signal transformation.³² Advancements in materials science and nanostructured design have significantly expanded the range of materials available for these applications, from traditional conductors like metals and metal oxides to innovative options such as carbon-based materials, conductive polymers, and semiconductor hybrids.^11,32–34 Highly conductive and intrinsically stretchable electrodes, crucial for various soft electronics, could be developed by integrating conducting nanomaterials with elastomers.³⁵ These materials are selected for their excellent electrical and mechanical properties, essential for delivering precise stimulation to tissues. However, challenges in achieving high conductivity, flexibility, biocompatibility, and biodegradability in functional materials remain, alongside the need for antibacterial properties and innovative features like self-healing and stimulus-responsive performance, which enable materials to adapt to physiological changes. The transition from materials to device development involves sophisticated engineering to create wearable or implantable devices that are effective yet user-friendly. Emerging device formats include soft electronic skins, injectable/implantable bioelectronics, and organically integrated bioelectrodes, designed to administer precise electrical doses tailored to individual therapeutic needs.^1,36–40 Despite the progress, issues such as materials' stability, device miniaturization, manufacturing scalability, and the integration of real-time monitoring and feedback mechanisms persist. Regulatory and ethical considerations also present substantial challenges, although advances in nanotechnology, microfabrication, and artificial intelligence offer promising avenues to address these hurdles and enhance the functionality and adaptability of ES based bioelectronics across various medical applications.

At the cellular level, ES impacts the transmembrane potential (TMP), where a vital voltage differential across the cell membrane is essential for biofunctions such as communication, growth, and metabolism.^41–43 This review starts with a comprehensive exploration of how ES alters the TMP of cells, significantly impacting cellular functions such as attachment, migration, alignment, proliferation, apoptosis and differentiation, as detailed in Section 2. ES can be applied in vitro to cells through several methods including direct coupling, capacitive coupling, and inductive coupling, each of which is characterized by distinct mechanisms and applications.^44,45 The modifications in cellular activities via electrical inputs offer critical insights for therapeutic applications and advancements in medical technology.

Building on these biological and electrical foundational concepts, bioelectronics engineered with optimal physicochemical properties can effectively deliver ES to cells, tissues, and organs, enhancing biofunctions and treating diseases in clinical settings. The outcomes of these interventions are contingent on finely tuned parameters of the stimulation, including intensity, waveforms, and duration, which must be carefully optimized to secure the desired therapeutic effects,¹⁹ which is elaborated in Section 3. The precise tuning of stimulation parameters is essential for targeting desired tissues without affecting the surrounding areas, which needs a profound understanding of the electrical properties of biological tissues and the interactions with electrodes. Continuous monitoring and adaptive control systems ensure therapeutic safety and effectiveness.

Section 4 catalogues a diverse array of materials utilized in bioelectronics for ES, including metals, metal oxides, carbon materials, conductive polymers, semiconductors, and their hybrids. The selection of materials is based on the specific needs of the application, ensuring that the bioelectronic systems not only possess excellent electrical and mechanical properties, but also feature good tissue adhesion, biocompatibility, anti-bacterial properties, biodegradability, and other vital characteristics such as breathability, swellability, and self-healing capabilities.^46–48

In Section 5, the focus is on ES based bioelectronics for tissue regeneration, merging biology and engineering to develop functional replacements for damaged tissues that mimic the architecture and function of healthy tissues. ES influences the in vivo cellular behaviors and functions of various types of cells (e.g. fibroblasts, osteoblasts, and cardiomyocytes), and thus plays an important role in skin healing, nerve regeneration, bone regeneration, and drug delivery.^8,49

Section 6 delves into neuromodulation via ES, a versatile medical technique that alters nerve activity to modulate neurological and psychiatric conditions, influencing neural circuits for symptom alleviation and functional restoration through both non-invasive and invasive methods. This approach not only provides therapeutic benefits for conditions unresponsive to traditional therapies, but also enhances the understanding of nerve stimulation mechanisms and patient outcomes.^50–53

Section 7 highlights ES as an indispensable tool for biofunctional applications, effectively bridging the gap between bioelectronic systems and biological functions by delivering targeted electrical impulses to regulate heart rhythms, control pain, reduce neurological tremors, and modulate sensory feedback.^54–56

Section 8 is the conclusion and outlook with a discussion on the future directions of bioelectronics for ES. This overview emphasizes the complexity of biological behaviors and clinical scenarios, underscoring that each application must adhere to its underlying physiopathological mechanisms and specific requirements to ensure efficacy and safety of ES.

2. Cellular responses to ES

Cellular responses to ES bridge the gap between bioelectrical phenomena and cellular functions. A TMP voltage difference exists in all living cells, which is regulated through a delicate balance between the accumulation of various ion concentrations inside and outside the cells.⁵⁷ Consequently, when applying an exogenous ES to a cell, its TMP is altered, which can remarkably impact the cell metabolism and behaviors like alignment, migration, proliferation, apoptosis, differentiation, attachment, etc.⁵⁸ Understanding how to manipulate cellular activities through electrical inputs not only offers fundamental knowledge for therapeutic applications, but also offers guidelines for the development of medical technologies. The influences of ES at molecular and cellular levels are illustrated in Fig. 1 and comprehensively summarized in the subsequent sections.

Fig. 1 The molecular and cellular-level changes activated by ES.

2.1. Effects of ES on cell behaviors

2.1.1. Cell alignment and migration. Guided cell alignment and migration play pivotal roles in the field of regenerative medicine since they directly participate in embryonic development, tissue formation/regeneration, and wound healing.^59,60 Recently, many strategies such as use of bioactive factors^61,62 and specific surface patterns^63,64 have been put forward to induce cell migration and alignment, of which ES has shown the maximum efficiency.^12,65 ES can exert a pronounced influence on cell alignment and effectively redirect the randomly oriented cells towards a desired alignment.^66,67 However, the direction of cell alignment in response to ES varies significantly depending on the specific cell types. For instance, various cell types including cardiac adipose tissue-derived progenitor cells, adipose-derived stromal cells, endothelial progenitor cells, vascular endothelial cells, and bone marrow stromal cells tend to align perpendicularly to the direction of the electric field. The fundamental mechanisms driving the alignment of these cells typically involve the activation of voltage-gated ion channels, G-protein coupled receptors, integrins, and cellular polarization processes, which facilitates a more uniform distribution of electric forces within the cells by minimizing the endogenous electric field gradients.^59,68–70 Conversely, ventricular myocytes, cardiomyocytes, myoblasts, and osteoblasts exhibit a distinctive tendency to align themselves parallel to the electric field vectors, which is mainly caused by the cytoskeletal rearrangement induced by ES.^71–74 In general, the intensity of applied ES was kept below 10 V cm⁻¹ in these studies. In addition, there is a trade-off between enhanced cell alignment and the potential impact on cellular functionality; although the increase of ES intensity facilitates cell alignment, cell activity tends to decrease correspondingly.^12,70

Likewise, ES affects the cell migration,⁷⁵ and its directional migration in response to ES is commonly referred to as electrotaxis.⁷⁶ It is crucial to acknowledge that the impact of ES on cell migration varies with the specific type of cell involved. For example, neural stem cells, macrophages, osteoblasts, and endothelial progenitor cells are generally observed to migrate towards the cathode,^70,77–79 while human dermal fibroblasts and Schwann cells tend to migrate towards the anode.^80,81 Interestingly, cells of similar types, but derived from different sources, exhibit distinct electro-kinetic properties. Taking mesenchymal stem cells (MSCs) as an example, adipose tissue-derived MSCs (AMSCs) have been found to demonstrate different traveling wave velocities and rotational speeds in comparison to bone marrow-derived MSCs (BMSCs).^82,83 Furthermore, it has been discovered that reversing the polarity of ES can also reverse the direction of cell migration.⁸⁴ Additionally, research has shown that ES intensities ranging from 0.1 V cm⁻¹ to 12 V cm⁻¹ could promote cell migration without causing substantial cellular damage, altering cell phenotype, or impairing differentiation potential.^85–87 Furthermore, it has been revealed that higher ES strength leads to a gradual increase in the cell migration rate and distance. Currently, the specific mechanisms underlying cell migration remain incompletely elucidated; however, it is widely acknowledged that endogenous electric fields, various membrane-localized proteins (e.g., ion channels, membrane transporters and receptors) and interactions with signal transduction pathways (e.g., Wnt/GSK3β/β-catenin and TGFβ1/ERK/NF-κB signaling pathways) play a pivotal role in orchestrating this intricate cellular process. For details of the influence of ES on the cell migration, refer to other excellent reviews.^65,76

2.1.2. Cell proliferation and apoptosis. Despite widespread evidence suggesting that ES primarily boosts cell proliferation without affecting the phenotype, particularly under continuous stimulation at intensities below 1 V cm⁻¹, contradictory findings exist. Some studies indicate that ES can either decrease or have no discernible impact on cell proliferation.⁸⁸ For instance, one study showed that applying 10 and 50 mV mm⁻¹ for 1 hour per day resulted in no significant changes in the proliferation of rat BMSCs and AMSCs.⁸⁹ However, another study reported that prolonged use of direct current (DC) ES significantly enhanced the proliferation of rat and human BMSCs.⁹⁰ The observed enhancement in MSC proliferation following ES may be attributed to an increase in the secretion of trophic factors (e.g., nerve growth factors), which accelerate the transition of cells into the S phase of the cell cycle.^91,92 Similarly, the effects of ES on cell apoptosis, which may often occur concurrently with cell proliferation,⁹³ remain uncertain.⁸⁸ For example, numerous studies have demonstrated that ES increases the rate of cell apoptosis by altering cellular membrane permeability, which in turn triggers calcium ion (Ca²⁺) influx, activates the ERK1/2 pathway, changes the expression levels of diverse intracellular molecules (e.g., p53 and Bcl-2 proteins).^94–96 However, conflicting findings have shown that ES can alleviate cell apoptosis by modulating pro- and anti-apoptotic protein expression (e.g., Bcl-2 and caspase-3) and activating multiple signaling cascades (e.g., mitogen-activated protein kinase and phosphoinositide 3 kinase-protein kinase B signaling pathways) within the target cells.^97–101 More details can be found in a review written by Love et al.⁸⁸ Altogether, given the complex interrelation of cell proliferation/apoptosis with cell type/origin, maturation stages, the culture microenvironment and the regimen of ES, it is imperative for future studies to meticulously control these variables and delve deeper into the effects of ES on the cellular processes.

2.1.3. Cell differentiation. ES-induced cell differentiation represents a highly promising strategy in the field of tissue engineering, as it facilitates the integration of cells with scaffolds ex vivo. This is beneficial for constructing three-dimensional (3D) in vitro tissue models and preparing in vivo tissue implants.^102,103 Generally, the application of ES for short durations (several minutes) at low intensities (0.06–6 V cm⁻¹) has been shown to effectively stimulate the differentiation of human induced pluripotent stem cells into cardiac cells and muscle precursor cells.¹⁰⁴ This process has been demonstrated to be associated with upregulation of the expression of critical cardiac transcription factors (e.g., nkx2.5 and tbx5) and cardiac contractile muscle proteins (e.g., ACTC1 and TNNT2).¹⁰⁵ Furthermore, under ES conditions (a stimulation intensity of <2 V cm⁻¹ sustained over 14 to 28 days), rat bone marrow- and adipose tissue-derived mesenchymal stem cells, whether cultured in two-dimensional (2D) or 3D environments, tend to differentiate into the osteogenic lineage rather than cartilage, which is characterized by increased levels of alkaline phosphatase and collagen type I.^106,107 Additionally, many studies have demonstrated that appropriately applied ES can stimulate the differentiation of neural progenitor cells,¹⁰⁸ neural precursor cells,¹⁰⁹ and neural stem cells¹¹⁰ into neurons rather than glial cells, when cultured in differential medium commonly supplemented with retinoic acid and various nerve growth factors.¹¹¹ This enhanced neuronal differentiation of neural stem cells has been correlated with increased intracellular Ca²⁺ levels and elevated expression of heat shock proteins, which facilitate the increased transcriptional activation of neuronal genes.^112–115 It should be mentioned that the administration of ES to pretreated stem cell-seeded tissue engineered scaffolds holds the potential to induce a prolonged pro-differentiation effect. This will be beneficial for maintenance of cell bioactivities and improvement of tissue repair, which has unveiled new possibilities for the treatment of Alzheimer's and Parkinson's diseases.^116,117

2.1.4. Cell attachment. Cell attachment has been recognized as an important factor that significantly impacts cellular behaviors and functions. The surface micro-topological morphology of 3D scaffolds offers anchorage for cells while creating a niche for their growth. In vitro ES can further improve cell attachment onto scaffolds, which has been described by many studies.^118–120 For instance, Jaatinen et al. have observed significant alterations of the mouse myoblast cell line in cell adhesion when exposed to pulsed monophasic currents.¹²¹ In addition, a representative study has demonstrated the cell attachment is influenced by the charge of the electrode surface, where a positively charged surface improves cell adhesion.¹²² This finding might be ascribed to the negative charge of the cell membrane. Another interesting study also demonstrated that under biphasic pulsed ES (an intensity of 50 mV and a frequency of 250 Hz), a 3D electrically active matrix could facilitate integration of electrodes with cells by enhancing cell attachment and differentiation.¹²³ Although specific mechanisms remain to be fully elucidated, experimental evidence suggests that galvanotactic responses, Ca²⁺ flux, integrins and collagen fiber rearrangement play a crucial role in the ES-induced enhancement of cell attachment, which will be discussed in the following section.⁴² With the rapidly expanding domain of biomaterials, conductive polymeric materials have been widely engineered to administer ES to cells and even tissues, with the objective of fostering tissue healing and reconstruction.¹²⁴

2.2. Molecular mechanisms of cellular responses to ES

The underlying mechanisms that govern the cellular responses to ES are currently the focus of ongoing research.^125,126 Here, we summarize some critical hypotheses or views and discuss concisely below (Fig. 2). (1) Activation of various signal transduction pathways: exposing cells to external ES leads to the asymmetric redistribution or diffusion of electrically charged receptors on the cell membrane, which further triggers the activation of multiple downstream signaling cascades. For example, ES can activate mitogen-activated protein kinase cascades to regulate the transcription of specific mRNAs, which subsequently induces the expression of extracellular signal-regulated kinase, p38 mitogen activated kinase, PI3 kinase, etc.^127,128 Depending on the specific cell types and stimulation parameters, these molecular events can elicit various cellular responses, including proliferation, differentiation, apoptosis, and other processes. (2) Ion flow: ES can accelerate influx of sodium ions (Na⁺) into the cell, accompanied by an outward flow of potassium ions (K⁺) by trans-membrane delivery, which causes changes in the direction of cellular lamellipodial protrusion, cell volume and membrane potential.^129,130 In addition, after ES, intracellular Ca²⁺ is increased by the flow of extracellular Ca²⁺ through plasma membrane ion channels or the release of internally stored Ca²⁺ within the endoplasmic reticulum.^118,131 Ca²⁺ transients enhance the effectiveness and precision of gene expression and intervene in numerous vital cellular activities like proliferation, differentiation, and apoptosis. (3) Redistribution of surface receptors, lipid rafts and actins: the applied electrical field induces electrostatic and electroosmotic forces to redistribute the charged receptors on the external surface of the cell membrane. Likewise, these forces can induce the asymmetric redistribution of lipid rafts across the cell membrane, leading to the polarization, coalescence, and segregation of membrane proteins, such as integrin and caveolin, to guide directional cell migration.^132–134 Furthermore, these forces can induce changes in cytoskeleton tension through the rearrangement of cytoskeletal filaments and the reallocation of actin formation.¹³⁵ (4) Enhanced ATP synthesis: DC ES, within the range of 10 to 1000 μA, has been demonstrated to direct the migration of protons towards the mitochondrial membrane-bound H1-ATPases, consequently promoting ATP generation.^136,137 Extensive documentations reveal that intracellular ATP facilitates the reorganization of the actin cytoskeleton and thus influences various cytoskeleton-associated cellular processes.^138,139 More detailed information about the cellular mechanisms in response to ES was provided by Leppik et al.¹²⁶ and Zhao et al.¹²⁹ Altogether, these putative cellular mechanisms are intricately intertwined within a complex but meticulously orchestrated network that transforms the external electrical signal into cellular responses through receptor-based cell signaling.

Fig. 2 Possible mechanisms of cellular responses to ES. (a) Activation of signal transduction pathways. (b) Ion flow. (c) Redistribution of surface receptors, lipid rafts and actins. (d) Enhanced ATP synthesis.

2.3. Conventional methods to deliver ES in vitro

In most in vitro studies, cells are cultivated in either 2D or 3D microenvironments, providing specific regimens of ES within purpose-built chambers. The conventional methods for ES delivery can be primarily classified into three categories: direct coupling, capacitive coupling, and inducive coupling by the electromagnetic field (Fig. 3).⁴⁴ Among them, direct coupling is the simplest and most user-friendly method by use of culture chambers equipped with metallic electrodes (Fig. 3a). Typically, such electrodes are designed with one end bent to facilitate their immersion in the cell-containing culture medium. The opposite ends of the electrodes are subsequently connected to a power source, allowing for delivery of ES to the cells. Recently, various metals have been widely adopted as the direct-coupling electrodes for ES like copper (Cu),¹⁴⁰ platinum (Pt),¹⁴¹ stainless steel,¹⁴² titanium nitride (TiN),¹⁴³etc. Nevertheless, the limitations are still notable, including metal corrosion, inferior biocompatibility, abrupt temperature rise and pH alteration, cytotoxic electrochemical byproducts arising from the interaction between electrodes and the culture medium.¹⁴⁴ To mitigate these problems, some studies have implemented a peristaltic pump within the culture plates, which periodically exchanges the culture medium and reduces the effects of the contaminated medium on cells.^145,146

Fig. 3 Typical ways to deliver ES for cell behavior regulation, including (a) direct coupling, (b) capacitive coupling, and (c) inducive coupling by utilization of an electromagnetic field.

Compared with direct coupling, capacitive coupling is biologically safer for ES.^147,148 As illustrated in Fig. 3b, in the set-up of capacitive coupling, two electrodes are placed at opposite ends to create a uniform electric field across a scaffold, where cells are cultured, situated between these electrodes. Due to indirect contact with the medium, this system prevents cells from being exposed to toxic electrochemical byproducts and pH fluctuations.¹⁴⁷ More importantly, the capacitive coupling system is non-invasive and obviates the need for use of a conductive scaffold to form an electron channel.^147,148 Despite these advantages, capacitive coupling has many drawbacks: (1) inadequate penetration/induction depth limiting the study area for ES in a single setting; (2) non-specific stimulation that may extend beyond the intended target area and affect a broader cell population, thus compromising the ES accuracy; (3) uneven electrical field distribution due to variations in electrode placement and electrical conductivity of the culture medium, which can lead to inconsistent stimulation levels, making it challenging to achieve consistent and reproducible results across experiments.

Inductive coupling is another method for in vitro ES, which commonly involves the utilization of a conductive coil positioned around the cell culture system to generate a controllable electromagnetic field for stimulation (Fig. 3c).⁴⁵ The stimulus is transmitted in a pulsatile manner to replicate the natural transfer of electrical potential observed within the human body. Instead of directly applying ES to cells, this pulsed electromagnetic field can provide potential in proximity to the target cell.^45,144 Consequently, inductive coupling systems for ES show similar advantages to capacitive coupling systems. However, the significant time and resource consumption are the primary shortcomings of electromagnetic field stimulation. For instance, some treatments require lengthy durations of up to 10 hours per day, and in some cases, high voltage is needed to achieve the desired effects of ES.^149,150

2.4. Novel platforms for manipulation of cell behaviors

The above ES devices necessitate an external power supply to connect electrodes and/or coils, for applying an electric field, which are restrained by their inherent drawbacks like bulky size, high cost, suboptimal performance, and adverse environmental effects. Moreover, they are unrealistic for in vivo use, since the patient will have to undergo another surgery operation to take out the device and replace the old battery with a new one, once the battery runs out of power. The potential leakage of the battery may also lead to concerns of biosafety, since most of the state-of-art batteries involve chemicals that have severe biotoxicity to the biological system. Thus, development of ES devices that possess the desirable attributes of miniaturization, efficiency, safety, and eco-sustainability is rapidly gaining significant attention. Recently, instead of using power supply, many studies have adopted some novel energy transmission and/or harvesting approaches for ES, such as triboelectric nanogenerators (TENGs),^151,152 piezoelectric nanogenerators (PENGs),^153,154 radio frequency (RF),^155–158 ultrasound,¹⁵⁹ light,^160,161 and magnetic induction.^162–164 For instance, Chu et al. prepared a TENG for current stimulation in vitro and in vivo and found that such a TENG could inhibit cancer cell migration and early tumor metastasis (Fig. 4a).¹⁵¹ In another study, Chen et al. used polyvinylidene fluoride foam incorporated with zeolitic imidazolate framework-8 to construct a PENG, whose voltage output reached up to 10 V without the occurrence of polarization. They demonstrated that this PENG could enhance the in vitro angiogenic activity of human umbilical vein endothelial cells and osteogenic potential of BMSCs (Fig. 4b).¹⁵³ More examples of novel platforms based on the abovementioned energy supply technologies can be found in following sections in this review and reviews elsewhere.¹⁶⁵

Fig. 4 Novel in vitro ES platforms to regulate cell behaviors. (a) A TENG-based platform for suppressing in vitro cancer cell migration and in vivo tumor metastasis.¹⁵¹ (i) Schematic illustrating the structure of a TENG-based platform. (ii) Schematic showing the current stimulation of a TENG-based platform to inhibit cancer cell migration. Reproduced from ref. 151 with permission from Elsevier, copyright 2022. (b) Development of a PENG-based platform to realize angiogenesis-osteogenesis coupling.¹⁵³ (i) Schematic illustration of the preparation process of a PENG-based platform. (ii) Schematic showing the piezoelectric principle. (iii) The piezoelectric output under 1 m s⁻² acceleration. (iv) Improved angiogenic effect and (v) osteogenic activity of this PENG-based platform in vitro. Reproduced from ref. 153 with permission from Elsevier, copyright 2023. (c) Preparation of a 3D printed and electrically active hydrogel for continuous ES of human BMSCs.¹⁶⁶ (i) Schematic diagram of the fabrication process. (ii) Schematic demonstrating the direct current stimulation to regulate stem cell proliferation and differentiation. (iii) Alizarin Red-S staining and quantitative analysis of the osteogenic differentiation ability of human BMSCs. Reproduced from ref. 166 with permission from Elsevier, copyright 2023. (d) Fabrication of a microfluidic platform to mimic wound healing.¹⁶⁷ (i) Microchannel design to achieve different electric field distributions. (ii) Schematic showing the process of microfluidic preparation and cell seeding. Reproduced from ref. 167 with permission from Royal Society of Chemistry, copyright 2023.

The conventional cell culture platforms employed to assess the effects of ES are predominantly 2D in nature, typically established by directly seeding cells onto culture dishes.¹⁶⁸ Despite the ease of use, they are unable to re-create the intricate 3D cell–cell and cell–matrix interactions that exist within the human body. As a result, their ability to accurately recapitulate the physiologically relevant structures and functionalities of normal tissue is limited.^168–170 To address these issues, many studies have developed 3D tissue models by culturing cells within or onto extracellular matrix-like materials, such as hydrogels,^166,171 electroactive polymers,^172,173 inorganic materials,¹⁷⁴ and their composites,^175–177 to better mimic the architectural and chemical intricacies observed in living tissues. For example, Dutta et al. fabricated a full-thickness bone model with high-resolution biomimetic architectures by bioprinting of a conductive polypyrrole (PPy)-grafted gelatin methacryloyl (GelMA) hydrogel (Fig. 4c).¹⁶⁶ Under continuous microcurrent stimulation (250 mV/20 min per day), this hydrogel could accelerate osteogenesis of human BMSCs without impairing cell variability. Despite their high efficiency in cell behavior manipulation, unfortunately, these 3D models still fall short of fully recapitulating certain essential characteristics inherent to native tissues, such as the skin barrier function and corticostriatal network.¹⁶⁸ Currently, there is an increasing number of studies that utilized microfluidic devices to fabricate 3D tissue models for ES, commonly known as “tissue-on-a-chip”.^178,179 These models offer precise control over cell shape/position, various physical and biochemical cues, and the 3D structural organization of target tissues. By combining with electrically stimulated electrodes, these models enable the exploration of cellular behaviors within a microenvironment that closely mimics physiological conditions.¹⁷⁸ Shaner et al. constructed an in vitro wound healing model by combining microfluidic design with conductive composite materials (laser-induced graphene and poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) hydrogel) (Fig. 4d).¹⁶⁷ This interesting model effectively emulated the standard scratch assay, confirming that unidirectional electric guidance cues were better for wound closure. Furthermore, it was revealed that unidirectional ES could even restore impaired mobility in diabetic-like keratinocytes. In another study, Kim et al. prepared a microfluidic chip by integrating a microelectrode array as an in vitro central nervous system (CNS) model to induce neurite outgrowth and neural cell differentiation.¹⁸⁰ Similarly, Li et al. also demonstrated that the microfluidic-microelectrode integrated microdevice facilitated the research on the ES influence on neural stem cells in vitro.¹⁸¹ Collectively, these microfluidic platforms show great potential as a valuable bridge between 2D culture systems and the in vivo environments, facilitating a more precise recapitulation of in vivo cellular behaviors and functionalities in response to ES.

To achieve the desired modulation of cell behaviors, in addition to the platforms for ES delivery, comprehensive consideration of various parameters related to ES is also required. These stimulation parameters often encompass intensity, duration, waveforms, and frequency,^44,182,183 which will be discussed in the next section.

3. Principles of effective ES

Effective ES in bioelectronics relies on several core principles that ensure both efficacy and safety in therapeutic applications. First, the specificity of the stimulation must be finely tuned to target the appropriate tissues or cells without affecting the surrounding areas. This requires a deep understanding of the electrical properties of various biological tissues and the dynamic interface of electrodes within them. Additionally, the parameters of the stimulation—such as intensity and duration, monophasic or biphasic stimulation, waveforms (e.g., pulse, sinusoidal, square, triangular, or sawtooth patterns), and intermittent or continuous stimulation—must be carefully optimized to elicit the desired biological responses while minimizing potential adverse effects. Lastly, continuous monitoring and adaptive control systems are essential to adjust the stimulation for guaranteeing the safety of the biological system, ensuring that the treatment remains within therapeutic thresholds and is responsive to the changing conditions of the target tissues.

3.1. Electrode–tissue interfaces

Different from cells cultured in vitro, all tissues such as muscles, neurons and skin are in a complex aqueous microenvironment containing water, electrolytes, and various kinds of physical and chemical cues. Thus, the electrode–tissue interface plays a vital role in efficient ES in vivo.

3.1.1. Stimulation electrode interface modelling. Taking a neurostimulator as an example, it could be viewed as an electrochemical cell that is tested using a potentiostat or galvanostat (Fig. 5a).¹⁸⁴ Stimulation electrodes, used for both recording and delivering currents, always function in pairs: one serves as the stimulating electrode at the target position, delivering the current, while the other receives the current.³³ These can be analogized to the working electrode (WE) and counter electrode (CE) in an electrochemical cell. Although tissues exhibit anisotropic properties that differ from those of a salt solution, the charge transfer characteristics occurring at the interface between the electrode and the electrolyte are remarkably similar.¹⁸⁵

Fig. 5 Stimulation electrode interface modelling. (a) Principle of stimulation with a neural interface. (b) The equivalent circuit model of the electrode–tissue interface and schematic of different charge injection mechanisms.¹⁸⁴ (c) A typical Bode plot for the EIS characterization of the interface, including the impedance and the phase versus frequency.¹⁸⁴ (d) The charge-balanced stimulation waveform, for characterizing CIC.¹⁸⁴ (e) The pulse-clamp method for characterizing CIC.¹⁸⁴ (f) A typical CV plot (within the water window) for the Pt electrode in saline.¹⁸⁶ Adapted from ref. 186 with permission from American Chemical Society, copyright 2008. (g) Region of charge and charge density where neural damage was observed.¹⁹ Adapted from ref. 19 with permission from Elsevier, copyright 2005. (h) The Pourbaix diagram of the Pt electrode in 30 mM H₂O₂/PBS solution, where the relationship between corrosion resistance, pH and electrode potential is visualized. SHE, standard hydrogen electrode.¹⁸⁴ (i) Typical weak spots of the electrode (red dashed circles) in different configurations.¹⁸⁴ Adapted from ref. 184 with permission from Springer Nature, copyright 2020.

Considering a pair of solid metal electrodes that are brought into contact with a liquid electrolyte solution, Boehler et al. built up an equivalent circuit model for describing the electricity pathway between these electrodes, as shown in Fig. 5b.¹⁸⁴ In this model, the double layer capacitance is depicted as C_WE and C_CE near each electrode. An electrical double layer (EDL) forms at the interface between the electron-conductive metal electrode and the ionic solution, and the charge interchange between them primarily relies on capacitive coupling.¹⁸⁷ The surface of the metal electrode acts as one plate of the capacitor, charged with electrons, while the surficial layer of ions in the solution, organized in a Helmholtz plane, acts as the second plate of the double-layer capacitor. Consequently, only alternating current can pass through the EDL via capacitive coupling. This is reflected in the natural high-pass filter-like performance observed from electrochemical impedance spectroscopy (EIS) results (Fig. 5c).¹⁸⁴ At higher frequencies, the impedance decreases, whereas at lower frequencies, the impedance increases. In the intermediate frequency range, there is a linear transition in impedance and a 90° phase shift. When the frequency exceeds the high-pass cutoff frequency, the capacitive impedance becomes neglectable and the frequency-independent access resistance (R_A) of the bulk electrolyte becomes dominant.¹⁸⁷ The capacitance of this configuration is proportional to the contact surface area, so a straightforward method to lower the impedance is to maximize the effective surface area within the electrode's size, for example, by increasing surface roughness.

As mentioned above, the interfacial impedance between bioelectronics and tissue is lower at higher frequencies, which can be beneficial for current transmission in both directions (from bioelectronics to tissue or vice versa). However, electrical biosignals typically fall within a lower frequency range (in the EIS image), generally below 1000 Hz. For example, electrocardiography (ECG) signals range from 4 to 500 Hz,^188,189 electromyography (EMG) signals from 0 to 500 Hz,¹⁹⁰ electroencephalography (EEG) signals from 0.1 to 1000 Hz,¹⁹¹ and gastric electrophysiology (electrogastrography, EGG) signals from 0.01 to 5 Hz.¹⁹² According to the Nyquist theorem, biosignal readings should be conducted with a sampling rate at least twice the maximum frequency of the signal of interest to sufficiently capture all details.¹⁹³ At higher or extremely low frequencies, interfering noises can couple into the sampling system through various mechanisms. These include baseline drifting (∼1 Hz) caused by motion-induced unstable interfacial impedance or electrochemical reaction potentials, and high-frequency noises from other biopotential signals (e.g., unwanted EMG signals in ECG readings), electromagnetic interference, RF interference, power line interference, and electrostatic discharge through capacitive, inductive, or radiative coupling. Thus, both high-frequency and DC noises should be filtered out to obtain a clear signal of interest. Additionally, phase shifts in the received signal at higher frequencies can cause significant distortion and delay, necessitating a lower sampling frequency for accurate and timely sensing.

Capacitive coupling is not the only contributor to the charge transfer. Direct faradaic currents that involve redox chemical reactions could also exist in parallel with the capacitors; thus, the interface won’t be totally insulative in the DC range.¹⁸⁴ When the applied voltage exceeds a certain electrochemical window, charge injection will be contributed by electrons combining with positive ions or negative ions losing electrons, which embodies as a faradaic impedance (Z_F). Since multiple electrochemical reactions may occur at the interface, Z_F could be composed of multiple parallelly connected impedances. Faradaic currents that cause electrode corrosion, changes in electrolyte composition, or gas generation at the interface should be avoided, especially in long-term implantable applications. Therefore, it is crucial to characterize the electrical performance of an electrode to achieve stimulation without involving faradaic currents. Furthermore, the frequencies of ES are primarily controlled by the properties and response of the target tissues or organs. Different ES waveforms can induce varying performances, which will be discussed in the following applications.

3.1.2. Essential electrical characteristics of ES interfaces. For evaluating the electrical performance of the electrodes at the stimulating interface, it is important to figure out what are the electrochemical activities happening there, and thus an in vitro experiment setup is needed. The immediate surrounding environment of the stimulating microelectrode could typically be simulated by phosphate buffered saline (PBS), which has many similarities to in vivo environments and contains electrolytic ions that are comparable to the extracellular fluid (ECF), such as Na⁺, Cl⁻, and K⁺, as well as a close pH value and conductivity, etc.¹⁹⁴ Although phosphate ions in PBS took the place of carbonate ions in the ECF and PBS doesn’t contain nutrients, metabolites and other organic substances, the fundamental electrochemical performance of an electrode won’t be much affected by these minor factors.¹⁹⁵ Thus, PBS could serve as an in vitro testing environment that simulates the in vivo conditions for short-term characterization. This electrochemical testing setup also involves three electrodes, the WE that represents the stimulating electrode, the CE as the ground or current sink, and the reference electrode (RE) that represents the standard potential. The accurate potential of the WE and CE could be measured and separately controlled according to the actual requirements.

As mentioned above, EIS is one of the basic characterization methods for evaluating the performance of an electrode. The basic principle is to analyze the voltage response to the sinusoidal excitation signal with frequency f in a vast range, typically 10⁰–10⁵ Hz, and a small amplitude of 10–100 mV, for avoiding the irreversible electrochemical reactions in the solution. By comparing the phase and amplitude of the recorded signal with the original input standard signal, the electrochemical impedance Z could be deduced, which is typically composed of the realistic part Re(Z) and the imaginary part Im(Z) as shown in eqn (1), and the total amplitude could be plotted on a logarithmic scale.

The measured spectrum also has the phase shift part plotted on a linear scale. These two parts comprise the Bode plot that reflects the transfer function of the interface, which means how smoothly charges could be injected into the solution, and how they change with varying frequency. With EIS results, the basic interfacial conductivity of an electrode could be analyzed.

For characterizing the charge injection performances, methods like the charge-balanced stimulation test or pulse-clamp test could be used.⁸ To know about the characterization mechanism, we firstly need to know the charge injection mechanism in the solid–liquid interface. The charge carriers in each phase differ, with electronic carriers present in the solid electrode and ionic carriers in the biological solution.¹⁸⁴ Thus, the injection of charges into a liquid can involve electrochemical reactions that are either reversible or irreversible. Irreversible reactions may result in alterations to the electrode material and changes in the electrolyte composition of the biological solution. Therefore, the primary aim of testing is to determine the threshold at which all reactions at the interface remain reversible. Usually, biphasic charge balanced pulse signals are used for ES (Fig. 5d),¹⁸⁴ where the first half of the phase (typically a high negative pulse for polarizing the cell membrane) is immediately compensated by the second half (a positive pulse with the same charge amount) by switching the polarity. The second half often has a longer duration and lower amplitude to reduce the necessary voltage excursion in the reverse phase. Therefore, the total amount of the charge transferred through the whole process remains zero. However, in addition to the EDL-dominated capacitive charge injection, which is entirely reversible, there also exists faradaic charge injection that leads to reactions at the interface.¹⁸⁵ This reaction could be either reversible or irreversible, especially when the products are volatile or diffuse away from the interface. Thus, although the charge is balanced, the products may not be perfectly restored to their original state. For fast alternating signals with voltage not exceeding a certain threshold, the dominating mechanism is non-faradaic capacitive coupling, but as the voltage exceeds the threshold, the proportion of faradaic current in total charge injection is increased.

As all biological tissues contain water, we can consider it as an aqueous environment. Thus, the most common irreversible faradaic reaction that happens in all electrode–tissue interfaces is the electrolysis of water.

2H₂O + 2e⁻ ↔ H₂ + 2OH⁻ (reduction)

2H₂O ↔ O₂ + 4H⁺ + 4e⁻ (oxidation)

The voltage thresholds of these reactions (both reduction and oxidation) define a range within them, which is called the “water window”.¹⁸⁶ Within the water window, the charge injected won’t cause the generation of volatile gas (hydrogen or oxygen) or changing of pH of the solution, and thus could be considered as a safe range for charge injection in ES. The term that best describes the charge injection performance of the electrode within the water window is called “maximum safe charge injection capacity” (CIC_max), which refers to the maximum charge of each pulse (either negative or positive) that can be injected within the water window.

Additionally, faradaic reactions could also contribute to the reversible charge injection.^196,197 When the products of reactions stick close to the interface and are less likely to diffuse, their local concentrations are not changed by diffusion and thus the electrochemical reaction is totally reversible. The best example is those happening in the interface of a sputtered iridium oxide (IrO_x) film electrode, where a series of reversible reduction/oxidation reactions take place within the porous film, as Ir forms and transforms among IrO, IrO₂, Ir₂O₃, Ir(OH)₃, Na[IrO(OH)₂ × H₂O], etc.

The highly reversible faradaic reactions also include those happening in conjugated polymers, where the electrostatic interactions between the electronically conductive conjugated backbone of the polymer and the ions that are absorbed in the bulk materials formed a huge faradaic pseudo-capacitance.¹⁹⁸ This implies that while faradaic reactions occur due to electron transfer, the charging and discharging behavior at the interface resembles that of a capacitor, where the voltage changes almost linearly with the transfer of charges and the capacitance value remains relatively constant. This is because the reversible faradaic reactions between the electrode and absorbed ions happen very rapidly. As a result, it appears that the absorbed ion is merely transferring charges without contributing to material or chemical changes. This pseudo-capacitance forms not only on the electrode surface but also throughout the total volume of the electrode.

Apart from the above-mentioned reversible faradaic reactions, other factors also exist that could affect the charge injection behavior whether reversible or irreversible. These factors include the temporal characteristics of the pulses (longer pulse reduces the current/voltage but is more contributed by faradaic reaction), the symmetry and frequency, the shape of the electrode (sharp edges and corners lead to higher electric field/current density), the electrolyte species, etc.

The CIC_max could be determined by a charge-balanced stimulation test as mentioned above. A current-controlled charge-balanced biphasic pulse is applied on the WE-CE circuit, where the CE is assumed to be much larger than the WE for avoiding the influence of the CE. By monitoring the voltage of the working electrode with an oscilloscope, we could observe the characteristics of each contributor of the current. As depicted in Fig. 5d, the terms that could be observed in the voltage drop include the initial polarization of the electrode V₀, the voltage drop over the electrode interfacial boundary V_WE, the voltage drop over the electrolyte (access resistance) V_A, and the electrolyte-concentration gradient-related concentration-over potential V_C.¹⁸⁴ The V₀ could be distinguished by the baseline of the signal and the V_A and V_C are rapidly reflected since they are constant or won’t change in a short time. Thus, any voltage changes in the constant-current period of one phase of the pulse could be easily recognized as the V_WE (V_WE-C for the cathodic phase and V_WE-A for the anodic phase). For reversible charge injection, the voltage during the whole stimulation period should be restricted within the water window. Thus, the maximum amplitude of the V_WE (either negative or positive) should be within the water window and the charges at a critical value could be used for identifying the CIC_max. This test typically starts from a relatively lower pulse current with fixed duration (e.g. 200 μs) and the current gradually increases until the V_WE becomes the tangent of the water window, and this threshold charge value could be defined as CIC_max.

Although the charge-balanced stimulation test could reveal CIC_max, there are still faradaic currents within the water window and their portion in the total charge transfer can’t be reflected in the test. To meet this need, another characterization method is proposed, which is called the pulse-clamp method.¹⁹⁹ This method requires swift switching between the current-controlled mode (galvanostatic) and the voltage-controlled mode (potentiostatic) for “clamping” the pulse after it is delivered. As depicted in Fig. 5e, a single cathodic square pulse is delivered with constant current with fixed duration (200 μs for mimicking neurostimulation) and is quickly changed to the voltage-fixed (to V₀) mode after a short delay period (∼50 μs) for monitoring the consequent current and charge changes.¹⁸⁴ By comparing the charges transferred during the pulse phase and the voltage-clamped phase, we can determine how many charges are quickly and fully recovered versus how many charges are slowly or cannot be recovered. Typically, the fast recoverable charge is attributed to EDL capacitive coupling, the slowly recoverable charge correlates to reversible faradaic reactions, and the unrecovered charge indicates irreversible faradaic reactions. The total charge of the threshold current where the unrecovered charge is exactly 0 could be defined as CIC_max. In this way, the composition of the CIC_max could be quantitatively analyzed.

For characterizing the electrochemical active area, experimental methods such as cyclic voltammetry (CV) could be used.²⁰⁰ The CV method is used to drive the electrode voltage (potential) to sweep between reducing and oxidizing potentials cyclically with a constant scanning rate (V s⁻¹) and record the corresponding current at each potential. In this method, the capacitive and faradaic currents both contribute to the current. The capacitive current follows , as C_EDL and are both constant, the capacitive current would contribute a static amplitude part to the total current, for both cathodic and anodic currents.¹⁸⁴ Typically, the dynamically changing current is related to faradaic reactions. For example, in the CV test of a Pt electrode immersed in PBS, the voltage within the water window (−0.95 and +1.2 V relative to Ag/AgCl) was swept (Fig. 5f). When the potential approaches the activation voltage of a reaction (V_r), the current shows a drastic increase, which is especially obvious when it comes to both sides of the water window, where hydrolysis happens.¹⁸⁶ If the kinetics drives the potential further away from equilibrium, the faradaic current will present an exponential growth. Within the water window, some current peaks could be observed. This occurs because as the potential approaches the V_r of a certain reaction, the reaction is activated and the current increases. However, the reactant is not unlimited and is rapidly consumed near the electrode. When the voltage exceeds the threshold potential, the reaction becomes diffusion-limited, resulting in a drop in the current. Therefore, the diffusion-limited faradaic surface interaction appears as a peak at a characteristic voltage. In cases where multiple faradaic reactions occur, there can be multiple peaks that may be individually recognizable or overlapping, making identification and analysis more difficult.

CV peaks could be used to analyze how the faradaic reactions change with varying parameters such as pH, concentrations, or product diffusion.²⁰¹ They could also be used to estimate the increase in the electrochemical active area from the peak area of the formation of a Pt–H bond on the Pt electrode surface.¹⁸⁵ Since the peak area is related to the charge amount in the reaction, it is proportional to the active surface area. Furthermore, the charge storage capacity (CSC) is determined from the total area enclosed by the complete CV sweeping curve, and it refers to the total amount of charges transferred no matter by what mechanisms. For neurostimulation, the most critical term is the cathodic CSC, which represents the area under the current = 0 line. This metric is a comparative performance measure for evaluating stimulating electrodes. It is noteworthy that only part of the CSC area is compatible with short pulse charge injection, since some electrodes have high pseudo-capacitance that enhances the CSC; however, only superficial layers could be used for stimulation but not the whole internal surface of the bulk.

Compared to traditional metal electrodes, conductive polymers, such as PEDOT:PSS, show a lower bulk conductivity than metals. However, due to their ionic/electronic mixed charge transportation mechanism, the electrochemical mismatch between the electrode and tissue is significantly reduced in conductive polymers. The conjugated π-stacking structure of PEDOT and the ionic conductivity of the polyanion PSS in PEDOT:PSS enhance both electronic and ionic conductivities, creating a porous structure that is highly permeable to ions. This results in low interfacial impedance and a non-faradaic or reversible faradaic current injection mechanism due to its large pseudocapacitance. When PEDOT:PSS is coated onto metal microelectrodes, the CIC increases significantly, up to 9.5 times that of gold (Au) (from 0.2 to 1.9 mC cm⁻²) and 3.2 times that of Pt (from 0.83 to 2.71 mC cm⁻²).²⁰² It also demonstrates a superior maximum current injection limit. Moreover, the superior flexibility, biocompatibility, biodegradability, and capability for surface functionalization make conductive polymers a better choice than brittle metal materials in many applications,²⁰³ which will be further discussed in the following sections.

3.1.3. Safety and stability guidelines. In addition to modulating performances of the electrode–tissue interface, some parameters of stimulation signals also play crucial role in safe and effective delivery of ES. For example, ES relies on pulses exceeding a certain threshold to activate the cell, tissue, or neural activities.²⁰⁴ Effective stimulation may also induce tissue damage, which should be avoided or minimized by appropriately selecting the waveforms and thresholds.²⁰⁵ Two widely accepted mechanisms explain tissue damage during stimulation. The first, mass action theory, attributes it to overstimulation causing changes in the local environment. The second involves toxic electrochemical reaction products exceeding physiological tolerance.¹⁹

Charge-balanced biphasic waveforms have been shown to reduce tissue damage compared to monophasic waveforms, highlighting the importance of managing toxic product generation. Physiological mass action, on the other hand, is linked to overstimulation of excitable tissue due to synchronous activity in large populations of neurons. Tissue damage results from ionic concentration changes, oxygen/glucose depletion, and excitotoxicity. Crucial factors include charge per phase and charge density, which determine neuronal damage during stimulation. Charge density integrates current over time, while charge per phase depends on current amplitude and pulse width. As charge per phase increases, safe stimulation charge density decreases.

The Shannon model of neuronal damage describes the relationship between charge density per phase (CDPP) and charge per phase (CPP) in determining the level of neuronal injury. According to Shannon's model, the logarithm of the CDPP is proportional to the logarithm of CPP raised to a constant power.²⁰⁵ Shannon's model is represented by the equation

log(D) = k − log(Q),

where D is the CDPP, Q is the CPP, and k is a constant that is derived experimentally.²⁰⁵ This model suggests that there is a critical threshold defined by k that separates stimulation parameters that lead to tissue damage from those that do not (Fig. 5g).²⁰⁶ This model is important in understanding the impact of ES on neural tissue and how different stimulation parameters can influence the extent of neuronal injury.

Electrodes can experience various failure modes depending on the type of material used, such as a metal or a polymer.²⁰² On one hand, metal electrodes, such as Pt, can be susceptible to corrosion in certain environments. The stability of metal electrodes can be analyzed using a Pourbaix diagram (Fig. 5h),¹⁹ which illustrates the corrosion resistance of the metal based on the pH value and electrode potential.¹⁸⁴ Understanding the Pourbaix diagram can provide insights into the conditions under which the electrode material is most stable and resistant to corrosion in different environments. The Pourbaix diagram helps in understanding the electrochemical behavior of metal electrodes by showing the regions where various reactions occur, such as the formation of oxides, hydroxides, or dissolved species. This information is crucial for predicting the performance and longevity of metal electrodes in different applications, including neural interfaces. On the other hand, polymer electrodes can fail due to hydrolytic and oxidative attacks on specific bonds. Hydrophilic polymers are vulnerable to water penetration, leading to gradual degradation.²⁰⁷ Additionally, oxidative attack reduces conductivity and insulation.²⁰⁸ Enzymatic attack is also a concern, linked to the immune system's response to foreign materials.

Electrode failure modes at the interface and substrate result from various mechanisms, including surface or bulk material degradation and issues related to adhesive bonding (Fig. 5i).¹⁸⁴ Delamination, a usual failure mode, arises due to insufficient adhesion and internal stress development over time.²⁰⁹ Probes often feature metallization and adhesion layers to promote performance in substrate adhesion, interconnecting metallization and electrochemical behavior with the electrolyte.²¹⁰ Additionally, corrosive reactions on the surface can lead to delamination, which occurs as the electrolyte infiltrates deeper structures via compromised insulation barriers. Volumetric changes in the electrode material, such as ion exchange, may also weaken the binding strength, potentially causing delamination, internal stress, and cracks.

Characterizing the stability of electrodes involves a comprehensive approach that considers various factors to ensure reliable performance over time. The deposition and manufacturing technology of electrodes and devices plays a crucial role in shaping their internal structure. These processes can introduce weaknesses that ultimately affect their longevity.¹⁸⁴ Taking a holistic perspective is essential for analyzing stability, as failure modes can be influenced by factors beyond just the electrode material, such as insufficiencies in adhesion or internal device failures. For instance, liquid creep between insulation layers due to water uptake through the substrate can lead to short circuits or internal corrosion, affecting electrode performance.²¹¹

To assess stability, it is essential to conduct aging experiments that simulate long-term usage in a shorter time frame.¹⁸⁴ Elevated temperatures can accelerate reactions, providing insights into material degradation over time. The Arrhenius equation describes how the rate constant of a chemical reaction varied with temperature. Specifically, the rate constant increases exponentially as the activation energy decreases. For most reactions, the rate doubles with every 10 °C increase in temperature.²¹² It can be used to estimate aging factors based on temperature changes, allowing for predictions of material behavior under different conditions. According to the Arrhenius law, the acceleration factor at 60 °C is roughly 5 compared to body temperature. To ensure accurate performance assessment, it is essential to validate accelerated aging conditions by including samples aged under normal conditions (e.g., PBS at 37 °C) and running them in parallel. This approach accurately reflects the material's behavior in real-world applications.

When assessing performance, high-resolution imaging techniques like scanning electron microscopy (SEM) can be utilized to verify stability and surface changes post-stimulation.²¹³ Additionally, complementary methods such as Fourier transform infrared spectroscopy (FTIR), energy dispersive spectroscopy (EDX), X-ray photoelectron spectroscopy (XPS), X-ray diffraction analysis (XRD), and electrochemical quartz crystal microbalance (EQCM) can be employed for in-depth analysis of dissolution, surface changes, and elemental composition, providing a more comprehensive assessment of electrode stability.¹⁸⁴ Considering the impact of electrolyte composition on electrode performance is pivotal, especially for porous electrode materials sensitive to changes in the supporting electrolyte. While PBS is commonly used for performance tests due to its relevance to the ionic composition of the extracellular fluid, complementary measurements in more complex electrolytes may be necessary for specific applications.

3.2. Stimulation parameters

The parameters of ES, including intensity, duration, waveforms, and frequency, require meticulous optimization to effectively trigger the desired biological responses. Careful calibration not only ensures the efficacy of the treatment but also minimizes any potential adverse effects, thereby enhancing both the safety and effectiveness of therapeutic interventions.

The cathodic monophasic waveform is highly effective for stimulation (Fig. 6a(i)). It involves unidirectional current pulses, avoiding reverse current flow.¹⁹ During monophasic stimulation, reversible processes (charging and discharging of the double layer capacitance) occur, while irreversible faradaic reactions produce free radicals that can damage myelin, cell membranes, and DNA.²¹⁴ However, due to the risk of tissue damage, monophasic pulses are typically avoided for long-term stimulation. In contrast, biphasic waveforms prevent tissue damage (Fig. 6a(ii)).¹⁹ The first phase achieves the desired effect, while the second phase reverses electrochemical processes. Charge balance in biphasic waveforms doesn’t guarantee electrochemical balance, and irreversible faradaic reactions can occur during both cathodic and anodic phases. A charge-imbalanced waveform reduces positive potentials during the anodic phase to prevent electrode corrosion (Fig. 6a(iii)). Incorporating an open circuit interphase delay between the stimulating and reversal phases enhances the stimulation threshold and prevents suppression (Fig. 6a(iv)). Rapid charge injection during anodic reversal quickly shifts the electrode potential away from the most negative range, minimizing tissue damage.

Fig. 6 Typical stimulation waveforms. (a) Pulsed waveform classification regarding the polarity, charge balance and interphase delay.¹⁹ Adapted from ref. 19 with permission from Elsevier, copyright 2005. (b) Waveform classification regarding the phase symmetry.

Another classification method of biphasic stimulation waveforms categorizes them into 3 different types according to their phasic symmetry.²⁰ Biphasic waveforms consist of cathodic and anodic phases, with parameters such as cathodic current (I_c), anodic current (I_a), cathodic half-phase period (t_c), anodic half-phase period (t_a), and interphase dwell (t_ip) varying with specific scenarios and electrode dimensions.

Biphasic symmetric waveforms (Fig. 6b(i)) consist of equal charge/phase for both positive and negative phases, ensuring charge balance of the waveform.²⁰ These waveforms typically have a leading cathodal phase, although anodal-first pulsing may be more efficient for activating certain neural elements.¹⁸ Charge balance in biphasic waveforms prevents irreversible reactions that can harm electrodes or the surrounding tissues. Biphasic asymmetric waveforms (Fig. 6b(ii)) also aim to achieve charge balance but with the current and pulse widths of each phase being slightly different.²⁰ The second phase operates at a lower current density and lasts longer than the initial phase, thereby minimizing positive polarization. This technique permits a higher positive bias during cathodal-first pulsing or a more negative bias during anodal-first pulsing. This approach aids in promoting charge injection while maintaining charge balance. Monophasic capacitor-coupled waveforms (Fig. 6b(iii)) involve injecting charge through a capacitor with a monophasic rectangular current pulse.²⁰ After the initial pulse, the capacitor discharges in the opposite direction via external switches that connect it across the stimulation and return electrodes. This method is commonly used with intramuscular and peripheral nerve electrodes, offering an alternative approach to charge-balanced stimulation waveforms.

4. Materials and their key performances for ES based bioelectronics

Metals, metal oxides, carbon materials, conductive polymers, semiconductors and their hybrid materials, etc. have been widely used for ES delivery.²¹⁵ These materials have their own distinct advantages, and thus the selection for ES should be based on the specific requirements of the intended applications. In principle, an ideal system for ES is supposed to possess excellent electrical performances, rigorous mechanical properties, good tissue adhesiveness, unique biological activities (e.g., biocompatibility and anti-bacterial efficacy), biodegradability, and some other key properties (e.g., appropriate breathability, swellability, and self-healing ability) (Fig. 7). In this section, we aim to present a comprehensive summary of the recent advancements in ES systems, focusing on their material compositions and key performances. Our objective is to provide readers with a comprehensive overview that will aid in the selection of appropriate materials and structure design for development of next-generation ES systems.

Fig. 7 Design principles for ES systems with desirable physicochemical and biological performances. Key properties for development of ES systems, including (a) electrical performances such as high electrical conductivity and high capacitance; (b) mechanical properties such as rigorous strength, high fracture toughness and stretchability; (c) adhesive performances such as high interfacial toughness (achieved by (i) chemical bonds; (ii) dynamic intermolecular interactions (e.g., hydrogen bonding and electrostatic interactions); (iii) physical entanglement of long polymeric chains; (iv) surface bionic microstructure-mediated adhesion); (d) biosafety and biological functions such as biocompatibility, anti-bacterial, anti-inflammatory, anti-fibrotic activities; (e) time-dependent biodegradable/bioresorbable properties; and (f) other properties such as breathability, waterproofness, and controllable swelling.

4.1. Electrical performances

Electrical performances are the key characteristics essential for the development of ES electrodes, which form the electrical communication from the device to the human body.²¹⁶ In Section 3.1.2, we have elaborated basic electrical parameters that determine the effects of ES such as EIS, CV, CIC, and CSC. In this section, we provide a summary of essential electrode materials, elucidating their electrical performances and the intricate interdependencies that exist among them (Fig. 8). Some typical examples of electroactive materials for ES applications are also listed in Table 1.

Fig. 8 Common electroactive materials as building blocks for ES systems.

Table 1 Recent and representative electroactive materials and their key features for ES applications

Types	Electroactive constituent	Material formulation	Electrical properties	Mechanical properties	Other key features	Biomedical applications	Ref.
Abbreviations: EIS, electrochemical impedance spectroscopy; CSC, charge storage capacity; Au, gold; Ag, silver; Ag NPs, silver nanoparticles; Pt, platinum; Ir, iridium; IrOx, iridium oxide; MnO2, manganese oxide; P3HT, poly(3-hexylthiophene); TiN, titanium nitride; PEDOT, poly(3,4-ethylenedioxythiophene); PSS, polystyrene sulfonate; PCL, polycaprolactone; PANi, polyaniline; PPy, polypyrrole; MA, methacryloyl; GelMA, gelatin methacryloyl; AlgMA, alginate methacryloyl; MoS2, molybdenum disulfide; TENG, triboelectric nanogenerator; PENG, piezoelectric nanogenerator; GO, graphene oxide; rGO, reduced graphene oxide; PLA, polylactic acid; CNTs, carbon nanotubes; PAGP, poly[(alanine ethyl ester)-(glycine ethyl ester)]phosphazene; BP, black phosphorus.
Metals	Au	Alumina-templated Au electrode	EIS: 4.6 kΩ at 1 Hz	—	Highly ordered 3D nanostructured electrodes	Neural interface	217
	Ag	Dopamine-modified HA/MA/Ag NPs hydrogel	Conductivity: ∼0.45 S m^−1	—	Anti-bacterial effect	Infected diabetic wound healing	218
	Pt	Pt-coated Pt/Ir wire	EIS: 83 ± 15 kΩ at 100 Hz	—	Nanostructured coatings with good process control and reproducibility	Neural interface	219
Metal oxides	IrOx	IrOx film-modified microelectrode	EIS: 23.6 ± 1.9 kΩ at 1 kHz	—	In situ pH sensing performance	Brain diseases and neuroscience	220
	MnO2	MnO2 nanoflower-integrated flexible optoelectronic device	EIS: 25 Ω cm^−2 at 1 kHz	—	Optoelectronic conversion	Neural interface	221
Semiconductors	P3HT	P3HT/PCL/PPy electrospun nanofibers	EIS: 80.30 ± 18.80 kΩ at 7 kHz	—	Optoelectronic conversion	Neural tissue engineering	222
	TiN	TiN nanowire- and/or thin film-based electrode	EIS: ∼265 Ω at 1 Hz for a nanowire electrode, ∼330 Ω at 1 Hz for a thin film electrode	—	Electrochemical stability	Neural interface	223
	MoS2	MoS2 nanosheets	—	—	Ultrasound-mediated ES	Brain diseases and neuroscience	224
		MoS2/gelatin MA hydrogel	Conductivity: 1.48 × 10^−9 S m^−1	Tensile stress of 42.33 kPa and elongation of 53.90%	TENG-based ES platform with photothermal effect	Diabetic wound healing	225
Conductive polymers	PEDOT	IrOx film- stabilized PEDOT/PSS electrode	CSC: 1200 mC cm^−2		Microfluidic-based DC ES platform	Electrotaxis research	226
		Alginate/PEDOT hydrogel	Conductivity: 4.60 × 10^−4 S cm^−1 at 1 Hz	Modulus: ∼25 kPa	Injectability	Neural tissue engineering	227
	PANi	Hydroxyethyl cellulose/soy protein isolate/PANi conduit	Conductivity: 1.7 S m^−1; resistance: 20.6 kΩ	—	3D sponge conduit	Neural tissue engineering	228
	PPy	PPy/silk fibroin conduit	Conductivity: 0.11446 ± 0.00145 S m^−1	Yield stress: 0.059 MPa	3D bioprinting and electrospinning-combined manufacturing process	Neural tissue engineering	102
Carbon-based materials	Graphene	Graphene incorporated collagen/PCL scaffold	Conductivity: 6.22 ± 0.59 S m^−1	Young's modulus: 32.02 ± 0.76 MPa; tensile strength: 3.70 ± 0.18 MPa; elongation: 47.31 ± 3.31%	3D electrospun nanofibrous conduit	Peripheral nerve regeneration	229
	GO	GO/cellulose scaffold	EIS: ∼6 kΩ at 1 Hz	—	Self-made ES device	Bone tissue engineering	230
	rGO	rGO fiber	Conductivity: 329.4 S m^−1	Young's modulus: 6.7 GPa; tensile strength: 192.6 MPa	3D fibrous conduit with aligned micro- and nano-channels	Peripheral nerve regeneration	231
		Polydopamine/rGO/GelMA hydrogel	EIS: ∼1000 Ω at 10 kHz	Young's modulus: 23.6 ± 1.2 kPa;	3D hydrogel scaffold	Cardiac tissue engineering	232
		PLA/rGO/PPy nanofiber	Conductivity: 1.46 × 10^−1 S cm^−1	—	3D electrospun nanofiber-based self-made ES device	Peripheral nerve regeneration	233
	CNTs	Polyethylene glycol-functionalized CNT/silk fibroin film	—	Tensile modulus: 21.16 ± 2.21 MPa; tensile strength: 3.11 ± 0.42 MPa	Optoelectronic conversion	Neural tissue engineering	234
		PAGP/CNT/polydopamine film	Conductivity: ∼0.002 S cm^−1	—	Film with flat surface	Bone tissue engineering	235
Others	MXene	MXene-based composite electrode	Conductivity: ∼55 S cm^−1; EIS: ∼205.6 ± 11.1 Ω at 1 kHz	Durability: ∼0.35 MPa for lateral shear test and ∼0.23 MPa for vertical pull test	Multiscale electrophysiology monitoring and stimulation	Neuromuscular tissue interface	236
		MXene film	Conductivity: 97.1 ± 1.6 S cm^−1	—	MXene-coated tissue culture polystyrene substrate	Neural interface	174
		Bacterial cellulose/MXene hydrogel	Conductivity: 7.04 × 10^−4 S cm^−1	Tensile modulus: ∼0.8 MPa; tensile strength: ∼0.25 MPa	3D porous scaffold	Wound healing	237
	BP	Polydopamine-modified BP/AlgMA hydrogel	Conductivity: ∼0.125 S m^−1	—	TENG and PENG-based ES platform	Bone tissue engineering	238
		BP/chitosan/gelatin hydrogel	EIS: ∼17 kΩ at 0.1 kHz; Conductivity: ∼ 0.289 S m^−1	—	Electrostatic induction effect-based wireless ES	Neural tissue engineering	47
		Hyaluronic acid-dopamine/Fe^3+/BP hydrogel	Conductivity: ∼0.3 S m^−1	—	pH sensitivity and anti-bacterial activity	Skin tissue engineering	239

---

Metals, especially noble metals like gold (Au), silver (Ag), platinum (Pt), and palladium (Pd), are the most common electrode materials for ES by virtue of their high electrical conductivity.^184,240 Moreover, various cutting-edge micro/nanofabrication techniques including magnetron sputtering, electrochemical deposition, etching, and photolithography, and 3D printing, among others can be utilized to manufacture ES electrodes with diverse prototypes such as Utah arrays, planar probes, and microwire arrays.²¹⁵ These metal electrodes are scalable with relatively low costs, making them highly suitable for customization to meet specific application requirements. Notably, metal nanoparticles like Au and Ag offer a unique combination of exceptional electrical and thermal conductivities typical of metals, augmented by the distinctive physicochemical properties inherent to nanomaterials. It is feasible to adjust their surface properties by surface modification with different polymers and/or ligands. Meanwhile, by coating and doping with other electroactive materials, the electrodes containing metal nanoparticles can achieve controllable conductivity and stiffness.

Some metal oxides such as IrO_x and manganese oxide (MnO₂) have emerged as promising materials for ES in recent decades owing to their high charge injection limit.^241,242 These metal oxides can mitigate the detrimental dissolution of metal ions by reversible redox reactions during charge injection and simultaneously retain high charge transfer capabilities.^243,244 For instance, IrO₂ exhibits a mixed conductivity and demonstrates reversible redox reactions involving the interconversion between Ir³⁺ and Ir⁴⁺. This unique property enables IrO₂ electrodes to achieve significantly higher charge injection capacities (0.5 to 8 mC cm⁻²), compared with conventional Pt and Pd electrodes (0.05–0.3 mC cm⁻²).^20,245,246 Various techniques such as sputtering and electrodeposition can be employed to fabricate IrO₂ electrodes with diverse surface morphologies that have undergone extensive in vivo investigations in the cervical vagus nerve, cortex, retina, basal ganglia nucleus, etc.^247–251

Semiconductors are solid materials classified based on conductivity or resistivity and energy bands in electronics. The conductivity of semiconductors falls within the intermediate range between that of an insulator and that of most metals.²⁵² Through incorporating electron acceptors (p-dopants) or electron donors (n-dopants) as impurities, their conductivity can be further improved.²⁵³ Recently, a molybdenum disulfide (MoS₂)-based floating-gate memory interdigital circuit, poly(3-hexylthiophene) (P3HT)-based nanofibers, and TiN-based film electrodes have been used for ES to regulate inflammation in tendon repair²⁵⁴ and neuronal differentiation and directed growth.^255–257 More fundamentals and designs about semiconductor-based electrodes have been comprehensively reviewed elsewhere.^252,258

Conducting polymers feature a distinct class of organic polymers with a conjugated backbone.^259,260 They possess outstanding electrical and optical characteristics comparable to those found in metals and semiconductors. Their electrical conductivity is attained through the oxidation of the polymer, resulting in the generation of positive charges along the polymer backbone. To maintain a charge-neutral state, negatively charged ions, commonly known as stabilizing counter ions, are doped into the polymer structure.^261,262 Due to narrow band gaps in conjugated polymers, electrons can transmit readily between the conducting band and the valence band. PEDOT,²⁶³ polyaniline (PANi),²⁶⁴ PPy¹⁰² and melanin^265–267 are the most common electrode materials for ES. Among them, PEDOT stands out as the most chemically and electrochemically stable conductive polymer, while possessing superior biocompatibility. Due to their high surface area and good electrical conductivity, PEDOT-based electrodes exhibit low electrical impedance and a high charge injection limit, making them highly favorable for ES applications. In addition, it is easy to deposit PEDOT onto various substrates by electro-polymerization of a bicyclic monomer, 3,4-ethylenedioxythiophene, and spin coating of a PEDOT mixture.²⁶⁸ However, the limited water solubility of PEDOT hinders its application in biomedical engineering, although the doping with polystyrene sulfonate (PSS) and tosylate anions can partially mitigate this drawback.^120,269,270 Notably, PEDOT:PSS is an intriguing conductive polymer with a unique combination of properties, offering mixed conductive mechanisms based on ionic and electronic transduction. This mixed conductive polymer can provide a smoother electron–ion transduction interface in bioelectronic electrodes. PEDOT:PSS contains both PEDOT (a conjugated polymer) and PSS (an ionic polymer) segments. The electronic conductivity is contributed by PEDOT, a conjugated polymer with alternating double bonds, due to its π-conjugated structure. The π-electron delocalization along the polymer backbone allows for efficient charge transport. Meanwhile, the ionic conductivity is contributed by PSS, a sulfonated polystyrene, where some of the sulfonyl groups in PSS are deprotonated, resulting in a negative ionic charge and thus could be ionic conductive. As PEDOT oligomers are polymerized onto long PSS chains, PEDOT:PSS exhibits a porous structure that is highly permeable to ions, allowing the bulk volume to contribute to electrochemically reactive areas. It also achieves high CSC and CIC without surface roughening, but rather by increasing thickness²⁷¹ or through solution treatment.²⁷² Rivnay et al. explored the role of morphology in the mixed ionic and electronic transportation performance and revealed that nano- and meso-scale structures in PEDOT:PSS films affected both bulk ionic and electronic mobilities (Fig. 9a).²⁷² The addition of co-solvent ethylene glycol (EG) to the casting dispersion has been shown to change the film architecture, potentially leading to PEDOT aggregation with tighter π-stacking. This phenomenon could result in changes in the size of gel-like particles, imparting a coarser and more heterogeneous morphology to the film and consequently impacting the effective ion mobility within the film. As the EG content increased, leading to larger domain sizes, a concurrent purification process occurred, impacting both the PEDOT:PSS-rich cores and the surrounding PSS matrix, consequently influencing the overall structure of the film. By quantifying domain composition, researchers determined an ideal morphology that supports balanced ionic and electronic transport, essential for the functionality of mixed conductor devices. In another study, to balance the electrical conductivity and modulus of the PEDOT:PSS-based hydrogel, Santhanam prepared a conductive hydrogel with an interpenetrating network by introduction of Ca²⁺-cross-linked alginate.²⁷³ These findings indicate that carefully choosing polymeric materials and refining processing methods can significantly enhance the performance of devices dependent on mixed conduction.

Fig. 9 Electroactive materials. (a) Mixed ionic/electronic conduction in PEDOT:PSS.²⁷² Reproduced from ref. 272 with permission from Springer Nature, copyright 2016. (b) High-performance graphene-fiber-based neural microelectrodes.²⁷⁴ Reproduced from ref. 274 with permission from John Wiley and Sons, copyright 2024. (c) Schematic diagram of an MXene that is synthesised from exemplary MAX phases, where M is an early transition metal (e.g., Ti, V, Cr, Zr, Nb, Mo, Hf, Ta) and X is carbon or nitrogen.²⁷⁵ Reproduced from ref. 275 with permission from John Wiley and Sons, copyright 2014. (d) MXene/PEG composite 3D printed scaffolds for synchronous beating of cardiomyocytes.²⁷⁶ Reproduced from ref. 276 with permission from Elsevier, copyright 2022. (e) MXene as a 2D filling material to increase tensile strength and modulus of nanocomposite membranes.²⁷⁷ Reproduced from ref. 277 with permission from Elsevier, copyright 2021. (f) Implantable McMiA assembly for reversible non-faradaic neurostimulation.²⁷⁸ Reproduced from ref. 278 with permission from American Chemical Society, copyright 2023.

PANi, a conducting polymer, has attracted significant research interest and exists in three distinct oxidation states: fully oxidized, semi-oxidized, and completely reduced. The half-oxidized form is particularly noted for its enhanced stability, conductivity, and ability to adjust its oxidation state, making it one of the most favorable for various applications.²⁷⁹ The major obstacle for PANi as the ES electrode is the potential cytotoxicity associated with its base and salt forms and chronic inflammation reaction following in vivo placement. PPy and its derivatives have also been extensively investigated as ES interface materials owing to their high conductivity, good biocompatibility, and stimulus-responsive redox properties.^117,280 Since the chemical composition of PPy comprises a rigid conjugated backbone (i.e., the recurring units of nitrogen-rich aromatic rings), it shows insolubility in water after polymerization of the pyrrole monomer. Likewise, the alteration in the doping conditions can regulate the conductivity of PPy, which allows for customization to meet the specific electrical needs.

Melanin is a natural polymeric pigment that possesses intrinsic conductive properties and potent anti-oxidation activity. In its chemical structure, there are complex heteropolymeric structures composed of 5,6-dihydroxyindole and 5,6-dihydroxyindole-2-carboxylic acid monomeric units, which can coalesce to form aggregated assemblies stabilized by strong π–π interactions between their aromatic moieties.^265,266 Thus, it is postulated that the presence of these extensive aromatic systems within the eumelanin structure provides the basis for the unique electrical conductivity and photoconductivity properties. Recently, melanin has been extensively incorporated into non-conductive polymer matrices (e.g., silk fibroin and poly-3-hydroxybutyrate) to impart the resulting composite materials with enhanced electrical conductivity properties.^267,281

Carbon-based materials, including graphene and its derivatives, carbon nanotubes (CNTs), and carbon fiber, have emerged as desirable conductive components for ES electrodes.^282,283 These materials possess captivating properties, including electrochemical stability, capacitive behavior, a broad electrochemical window, fast electron transfer kinetics, chemical inertness, and favorable biocompatibility. Taking graphene as an example, it is a 2D monolayer material with a hexagonal lattice structure, and exhibits exceptional physicochemical properties like large surface area, easy functionalization, adjustable size, excellent mechanical flexibility, and high electrical conductivity (a carrier mobility of ∼10 000 cm⁻² s⁻¹).²⁸⁴ Recently, graphene, graphene oxide (GO), and reduced GO (rGO) have been implemented as bulk conductors, coatings, and dopants for development of electroactive scaffolds for ES. For instance, Wang et al. put forward a method for fabricating high-performance graphene fibers as neural microelectrodes (Fig. 9b).²⁷⁴ These microelectrodes coated with a thin Pt layer showed high surface area, low impedance, and exceptional electrochemical properties. Their CIC reached 10.34 mC cm⁻² with an expanded CV window and a high signal-to-noise ratio, thus facilitating the measurement of neuronal activity. Such high CIC was achieved through the combination of a highly conductive Pt coating with a high-surface-area GO electrode, which enables efficient diffusion of electrolyte ions and facilitates ready electron transfer through the Pt layer. CNTs, another distinct category of carbon-based materials, possess cylindrical carbon structures, of which single-walled carbon nanotubes (SWCNTs) consist of a single carbon sheet layer and multi-walled carbon nanotubes (MWCNTs) comprise multiple layers.²⁸⁵ The unique structure bestows CNTs with attractive anisotropic conductivity (SWCNTs: 0.01 to 100 mS m⁻¹; MWCNTs: 0.1 to 10 mS m⁻¹) and high mechanical strength (a tensile strength of 89 GPa, a modulus up to 46 GPa, and an elasticity of ∼3.67%).²⁸⁶ It is noteworthy that the electrical characteristics of CNTs can be adjusted from metallic to semiconductive by manipulating the orientation of the hexagonal carbon lattice within the nanotube.²⁸⁷ Given the challenges associated with preparing long CNTs, they are commonly used as nanofillers of composites to fabricate ES electrodes with enhanced electrical conductivities and mechanical performances. However, although CNT-embedded composite matrices can realize high stretchability and flexibility, the concurrent rise in resistivity during stretching substantially impacts their electrical stability.

MXenes, 2D transition metal carbide nanomaterials, have gained significant attention in recent years as highly promising materials for ES due to their good electrical conductivity (∼1000 S cm⁻¹), high volumetric capacitance (∼1500 F cm⁻³), superior mechanical properties (a modulus of 330 ± 30 GPa for the monolayer titanium carbide MXene), and remarkable optical performance.^288–290 The exceptional electrical properties of MXenes are attributed to the presence of transition metals such as Ti, V, Cr, Zr, Nb, Mo, Hf, and Ta, the overlapping atomic orbitals within MXene layers, and their stacked layered configuration (Fig. 9c).²⁷⁵ Many studies have incorporated MXenes into scaffolds to enhance the electrical conductivity and mechanical properties of polymer matrices.²⁹¹ For example, Basara et al. printed polyethylene glycol (PEG)/MXene composite scaffolds to improve the synchronous beating and conduction velocity of cultured cardiomyocytes, achieving a higher conductivity than that of pure PEG (1.1 × 10⁴ S m⁻¹versus 0.1 S m⁻¹, respectively) (Fig. 9d).²⁷⁶ In another study, Fu et al. demonstrated that the addition of an appropriate quantity of MXenes could enhance the mechanical performance of nanocomposite membranes (Fig. 9e).²⁷⁷ MXenes also boast captivating optical properties, including strong plasmonic resonances, a broad optical transparency window, high optical transparency, and desirable photothermal conversion capabilities. These characteristics make MXenes highly promising for diverse biomedical applications such as photoacoustic imaging, photothermal therapy, and photodynamic therapy, as comprehensively reviewed elsewhere.^292,293 By introducing diverse functional groups onto the surface of MXenes, their surface topography, electronic structure, and surface charge can be readily modulated, further improving their stability and enhancing their electrical and optical properties. Most importantly, due to the presence of surface termination groups, MXenes can disperse uniformly in both aqueous and organic solvents, making them feasible to fabricate robust MXene-based thin-film electrodes for multiscale ES.²⁹⁴

BP is another conductive dopant that has attracted much scientific interest recently.²⁹⁵ Due to its distinctive 2D planar structure (i.e., a bilayer arrangement along the zigzag direction and a puckered structure along the armchair direction), BP possesses exceptional electrical conductivity, optical properties, and mechanical properties, which distinguish it from other 2D materials like graphene or MoS₂.^296,297 The investigation of BP-based ES platforms is currently in its infancy, and more advances are expected in the near future.^{238,298–300}

Apart from the abovementioned electroactive materials, hybrid conducive materials have also been proposed as the ES interfaces. For instance, it is possible to realize totally reversible non-faradaic stimulation with lower surface impedance, higher CIC, and a wider electrochemical window. This means that no chemical reaction occurring at the electrode–tissue interface is desirable, which requires a completely capacitive coupling between a solid electronic conductive electrode and the water-containing ionic conductive tissues. Kim et al. developed a hybrid multicross-linked membrane-ionogel assembly (McMiA) for reversible non-faradaic neurostimulation (Fig. 9f).²⁷⁸ The McMiA offered a buffering neural interface between neural electrodes and biological tissues, which possessed a wider electrochemical window than the extracellular fluid in natural biological environments for non-faradaic charge injection. The McMiA consisted of a genipin-cross-linked biopolymeric ionogel (X-BI) coupled with a dopamine-cross-linked GO (X-GO) membrane to prevent irreversible charge transfer reactions at the electrode–tissue interface. The X-BI contained choline-derived ionic liquids (ILs) and chitosan biopolymers, which made it both biocompatible and ionic conductive. The X-GO membrane was applied on the surface of X-BI, functioning as an ion diffusion barrier in the neural interface owing to its tortuous diffusion pathway of the stacked GO layers. The incorporation of biocompatible crosslinkers, such as genipin and dopamine, into X-BI and X-GO, effectively reduced their degradation resulting from hydrolysis, redispersion, or swelling. Dopamine that actively formed covalent bonds with biological tissue also endowed McMiA with bioadhesive and conformal nature to lower the impedance. As a result, McMiA presented an electrochemical window of 2.44 V, which was much wider than that in PBS (1.73 V). The observed phenomenon could be attributed to the increased concentration of “bound water,” which encompassed hydrophilic [Ch]⁺[MA]⁻ ions and water molecules, relative to the “free water” that was absorbed on the electrode surface. The increased presence of bound water played a crucial role in mitigating water electrolysis. By utilizing the developed McMiA, the in vivo stimulation demonstrated a notable increase of ∼75% in the volume of urine per voiding and an ∼35% extension in the interval between urination. The mean peak bladder pressure also showed a 17% increase after stimulation.

To summarize, these materials exhibit favorable electrical properties for ES through appropriate modification and processing. However, practical applications require consideration of additional characteristics such as mechanical and biological properties, which will be comprehensively discussed in the subsequent sections.

4.2. Mechanical properties

ES electrodes should have rigorous mechanical performances according to their specific applications. For example, as on-skin electrodes, they need to be soft and flexible to conformally contact the skin surface.^301,302 For in vivo probe electrodes for deep brain stimulation, it is crucial to possess a certain level of mechanical strength to facilitate penetration into the brain, while flexibility is also essential to reduce incidental tissue damage caused by brain movement. In general, achieving a harmonious match between the mechanical stiffness (or Young's modulus) of ES electrodes and the target tissues or organs is an essential requisite; otherwise, it may substantially diminish the efficacy of ES and lead to some adverse tissue responses, including perpetual generation of proinflammatory cytokines, fibrosis, and scar formation. In addition, the mechanical microenvironment inherent to the material itself has been observed to induce differential cell fates, which may potentially intervene in the outcome of electrode stimulation.³⁰³ Furthermore, because of the dynamic nature of the human body, high fracture toughness is highly desirable for the placed electrodes to withstand such dynamic mechanical loads and deformations.

While the material composition of an ES electrode predominantly determines its mechanical properties, numerous studies have demonstrated that meticulous geometric design can effectively complement and enhance the specific strengths of the materials employed. For example, bulk metals generally exhibit low fracture strain and high modulus. To augment the flexibility and stretchability of metal electrodes, thin-film technology has been utilized to prepare electrodes with serpentine or meander configurations onto stretchable substrates.³⁰⁴ In a representative example by Xu et al., a metal bilayer containing Au (5 nm) and Cr (200 nm) was sputtered in the form of serpentine pattern onto a silicone substrate (Fig. 10a).³⁰⁵ The whole device exhibited a Young's modulus of ∼69 kPa with the thickness of ∼20 μm, showing excellent conformality onto skin. Owing to the distinctive geometry, the maximum strain of the electrode was below 0.3% when the device underwent a tensile deformation of 21%. In addition, rational structural designs like the use of a tattoo-like geometry and a tripolar concentric ring can also achieve high stretchability and bendability.^306,307 However, as thinner and more flexible substrates with smaller feature sizes are employed, the compromised strength of the device becomes an escalating concern. One effective approach to address this issue is to introduce controlled defects like tears or holes within thin-film metal traces to facilitate a more uniform distribution of stress, thereby enabling the device to endure higher levels of strain before failure.³⁰⁸ Furthermore, addition of conductive materials like carbon-based materials into polymer matrices with high mechanical strength is another feasible approach to simultaneously orchestrate the mechanical and electrical performances of the resultant electrodes. For instance, Zhao et al. fabricated a conductive composite ES platform consisting of PPy and silk fibroin, where PPy is incorporated as the conductive component while silk fibroin is for enhanced mechanical strength.¹⁰²

Fig. 10 Mechanical properties. (a) A multilayer Au/Cr-PI construction showing excellent stretchability and compressibility.³⁰⁵ Reproduced from ref. 305 with permission from John Wiley and Sons, copyright 2023. (b) Soft strain-insensitive bioelectronics featuring brittle materials involve underlying electron transfer and out-of-plane and in-plane electron transport processes.³⁰⁹ Reproduced from ref. 309 with permission from AAAS, copyright 2022.

However, these geometric designs as well as addition of different components with high mechanical performances solely enhanced the mechanical stretching of electrodes, yet they did not address the issues of electrical property variations induced by tensile deformation. To make the stimulation performances of metal electrodes, especially intrinsically brittle noble metals (Au, Pt, or IrO_x), consistent at various strain levels, Zhao et al. proposed a strain-insensitive bioelectronic (SIB) architecture that allows the superficial brittle metal film to stay steadily interconnected even though it is cracked into multiple pieces after stretching.⁹ The architecture they used makes use of the out-of-plane vertical interconnections using anisotropically conductive film (ACF) (Fig. 10b).³⁰⁹ Through vertical interconnections within ACF, metal films are all connected to the underlying layer of stretchable silver nanowires (AgNWs), which function as a stretchable electron pathway network to allow all pieces of superficial metal to be still connected together under severe strain. In this manner, the total effective electrochemical active area of the metal electrode remains unchanged, as well as CSC and CIC. The neurostimulation performance of the IrO_x-SIB bioelectrodes was found to be highly effective and robust under deformation, delivering stable current and inducing muscle contractions with similar signal-to-noise ratios, as conventional electrodes even when strained. The studies involving sciatic nerve stimulation demonstrated efficient recruitment of motor units at low potential levels, with synchronized muscle contraction responses to varying stimulation frequencies. Furthermore, the c-Fos protein expression in the spinal cord indicated successful CNS neuromodulation in response to the sciatic nerve stimulation, highlighting the fidelity and effectiveness of the neurostimulation provided by the IrO_x-SIB bioelectrodes.

It is worth mentioning that many studies have adopted finite element analysis to evaluate the distribution of interfacial mechanical strains, aiming to direct rational structural design.^310–312 The current consensus is simple and uniform geometries enable more uniform distribution of strains across the entire device, which avoids concentrated strains along sharp edges.³¹³

4.3. Adhesion performance

ES electrodes require robust interfacial integration with the target biological tissues to deliver stimulation signals reliably, which necessitates high interfacial toughness.^216,314,315 Currently, three common strategies are utilized to achieve adhesion between electrodes and tissues, including the direct use of biocompatible adhesives, the construction of microstructures on electrodes to create a “dry adhesive” interface, and the combination of electrodes with pressure-sensitive adhesives.³¹⁶ In the first strategy, biocompatible commercial adhesives (such as water-soluble tapes,³¹⁷ silicone adhesives,³¹⁸ medical liquid tapes,³¹⁹ and acrylic tapes³²⁰) and the adhesives²⁶ developed in laboratory settings are applied between the electrodes and the tissues or directly onto the electrode surface as an encapsulation. For example, Yang et al. developed a new photocurable bioresorbable hydrogel adhesive as a heart-device interface for ES (Fig. 11a).²⁸ With strong physical and chemical bonding, the integrated device could conformally contact and stably adhere to the dynamic heart. The adhesion energy to epicardium reached 240 ± 20 J m⁻². Combined with the superior electrical performances, in vivo experiments demonstrated that this system can effectively provide ES to regulate the heart rhythm of rats for up to 8 days after surgery.

Fig. 11 Adhesion and biological properties. (a) (i) A photocurable hydrogel-based bioelectronic interface that (ii) could strongly bind to heart tissue and the electrode by complex physical and chemical interactions, enabling effective ES for over 8 days.²⁸ Reproduced from ref. 28 with permission from Springer Nature, copyright 2021. (b) (i) A adhesive bioelectronic interface integrated with a micro-channel network (inspired by tree frogs) and the convex structure (inspired by octopuses). (ii) Via capillarity-assisted suction stress and hydrogel absorption/swelling, such interfaces could (iii) form strong adhesion to the sciatic nerve and brain surface.³²¹ Reproduced from ref. 321 with permission from John Wiley and Sons, copyright 2022. (c) Selective peripheral neurostimulation is achieved through electrochemical modulation. Cuff electrodes are implanted on the sciatic nerve, inducing voltage-gated sodium channels and their interaction with extracellular Ca²⁺.³²² Reproduced with permission from CC BY 4.0 open access license, copyright 2022. (d) An ES electrode should possess critical biological functions, including (i) cytocompatibility,³²³ (ii) anti-bacterial³²⁴ and (iii) anti-inflammatory activities,³²⁵ as well as (iv) anti-fibrosis ability.³²⁶ Reproduced from ref. 323 with permission from Springer Nature, copyright 2024. Reproduced from ref. 324 with permission from American Chemical Society, copyright 2023. Reproduced from ref. 325 with permission from Elsevier, copyright 2021. Reproduced from ref. 326 with permission from Elsevier, copyright 2023.

The second strategy to enhance interfacial integration with tissues involves the fabrication of diverse microstructures on electrodes, taking inspiration from natural structures such as gecko feet,^327,328 octopus suckers,³²⁹ and clingfish discs.^330,331 For instance, Min et al. designed a CNT/P3HT-based conductive stretchable electrode with octopus sucker-inspired microstructures, which could simultaneously achieve adhesion and conductivity.³³² The incorporation of suction cups enabled a strong attachment to wet skin, exhibiting significantly greater adhesive force compared to unpatterned flat films. In another study, Kim et al. prepared a hybrid structure that combined the micro-channel network inspired by tree frogs with the convex structure inspired by octopuses (Fig. 11b).³²¹ This innovative design exhibited a substantial enhancement in tissue adhesiveness through an enhanced suction effect since the interfacial water could be drained effectively out by microchannels. Although these bioinspired microstructure arrays demonstrate robust adaptation to uneven skin surfaces under both dry and wet conditions, their complex fabrication process poses a challenge that needs to be addressed for a broader range of applications.

Another promising strategy involves the chemical modification of the substrates or electrodes to enhance their tackiness, typically forming pressure-sensitive adhesives. Similar to the abovementioned microstructure-induced “dry adhesive”, pressure-sensitive adhesives do not rely on chemical bonding for adhesion. Instead, their adhesion is derived from the viscoelastic energy dissipation of polymeric materials.³³³ As such, pressure-sensitive adhesives are able to adhere promptly to a diverse array of surfaces (like the skin or organ surface), upon the application of light pressure. This attribute makes them highly desirable for application as on-skin electrodes. In a recent study, Wu et al. prepared a polydimethylsiloxane (PDMS)-based pressure-sensitive adhesive with gradient cross-linking density by tailoring the diffusion process of a Pt catalyst.³³⁴ By controlling the concentrations of aminoethyl aminopropyltrimethoxysilane and its diffusion time into PDMS, this pressure-sensitive adhesive could achieve a peeling strength of 19.38 N m⁻¹, significantly greater than that of commercial Scotch tape.

Controllable tissue adhesiveness is of paramount importance in the development of bioelectronic interfaces.³³⁵ While strong adhesion is desirable during device use, it can potentially lead to discomfort or secondary injury when detaching from the skin after use. Hence, it is essential to strike a balance that allows for secure adhesion during operation while ensuring gentle and pain-free removal to minimize any potential adverse effects. To date, many researchers have showcased the capability of regulating interfacial adhesion by controlling the environmental conditions, including temperature,³³⁶ humidity,³³⁷ voltage,³³⁸ light,³³⁹etc.

4.4. Biosafety and biological functions

Biosafety is a critical consideration for in vivo ES, which is determined by both stimulation parameters³⁴⁰ and device components.³⁴¹ As mentioned in Section 3.2.2, ES with charge imbalanced waveforms, excessive interface polarization, uncontrollable current pulsing, and high access voltage have been shown to induce various irreversible chemical reactions.³⁴² These reactions can locally disrupt the in vivo microenvironment (e.g., increase the generation of reactive oxygen species) and potentially result in damage to both the electrode and the adjacent tissue. To prevent excessive voltage that activates these electrochemical reactions and increase the selectivity of stimulation, Flavin et al. implemented a lower stimulation voltage threshold by using an ion-selective membrane (ISM) for electrochemical modulation.⁸ By combining ES with focused Ca²⁺ depletion, the study aimed to selectively activate nerve fibers (Fig. 11c).³²² The authors developed an implantable ISM cuff and demonstrated in vivo that Ca²⁺ depletion increased nerve sensitivity to ES, selectively influencing functional components of the nerve. These findings suggested the potential for enhancing the functional selectivity of bioelectronic therapies. The study addressed challenges related to the selective activation of bundled nerve fibers and explored a novel approach to enhance stimulation selectivity through electrochemical modulation. Nevertheless, the biosafety assessment of a stimulation protocol, particularly in the realm of in vivo applications, lacks a universally recognized framework or definitive guidelines. More details have been comprehensively discussed in other reviews.^342,343

The electrode material itself can also affect the compatibility with targeted tissues. For instance, the direct and prolonged exposure of metal film electrodes to the skin can potentially result in skin irritation and allergic reactions.³¹⁶ For in vivo ES electrodes, they, as long-term implants, may induce various chronic tissue injuries, such as biofouling formation, infections, foreign body responses and the resultant fibrosis.²¹⁶ In addition, continuous corrosion of electrode materials when exposed to the biological tissues (because of dielectric cracking, delamination, and electrode site dissolution) presents an additional safety concern, as the by-products like toxic metal ions can increase oxidative stress and tissue inflammation.^344,345 This highlights the critical significance of electrical and chemical stability of electrode materials, besides the biocompatibility.

To address these biosafety concerns, a feasible approach is to integrate multiple biological functionalities into the design of electrodes, which allows for active regulation of various biological processes, including anti-infection, anti-inflammation, and anti-fibrosis. For example, many studies currently favor the incorporation of anti-bacterial components like Ag nanoparticles²⁶ and Ag/Zn coatings³⁴⁶ into ES electrodes. In some studies, various anti-inflammatory agents including human-like collagen,³⁴⁷ and tannic acid³²⁵ have also used to construct hydrogel-based electrode interfaces for anti-inflammation purposes. An old but representative study has demonstrated the direct release of anti-inflammatory steroids from the electrode site, which could significantly improve the efficacy of cardiac pacing electrodes.³⁴⁸ To minimize biofouling, the electrode surface is modified with various hydrogel and polymer coatings to enhance hydrophilicity or blended with molecules or drugs for anti-biofouling properties.³⁴⁹ Hydrophilic surfaces are able to establish a hydration layer through strong bonding with water molecules. As a result, they can serve as a physical barrier, effectively impeding the adhesion of cells and biomolecules. Common hydrophilic polymers against biofouling include PEG, PEG methacrylate (PEGMA) and poly(2-hydroxyethyl methacrylate) (PHEMA).³⁵⁰ Moreover, some zwitterionic polymers, such as poly(phosphatidylcholine), poly(sulfobetaine) and poly(carboxybetaine), are also widely utilized to modify electrode surfaces to form anti-biofouling interfaces.^326,351 By integrating these intriguing biofunctions, the electrodes can mitigate potential risks and promote a safer and more favorable physiological response within the biological system, thus improving the in vivo long-term performances of ES interfaces (Fig. 11d).^26,325,326

4.5. Biodegradable/bioresorbable properties

Traditional implanted ES devices require surgical removal after accomplishing their therapeutical purpose. However, secondary removal surgery not only adds an additional financial burden but also carries the risks of infections, surrounding tissue damage, and immune-mediated inflammatory reactions. To address these issues, bioresorbable or biodegradable materials have been used to construct the transient electronic systems.^352,353 These electronic devices gradually dissolve into the biofluid over time and ultimately generate biocompatible end-products.

The in vivo degradation of bioresorbable polymers involves several representative mechanisms that are of considerable importance in the controlled dissolution within the biological environment. These mechanisms include dissolution, enzymatic reaction, hydrolysis, and oxidative reaction (Fig. 12a).³⁵² Dissolution is a fundamental mechanism in which biofluids, primarily the water molecules, infiltrate into the network of solid polymer materials and lead to matrix swelling. Subsequently, polymer chains are desorbed and dispersed into the surrounding biofluid. Enzymatic reaction involves the interaction of enzymes in the surrounding biofluids with the polymer, resulting in the cleavage of chemical bonds in the polymer chains. Hydrolysis, another important mechanism, occurs when water molecules react with the polymer, leading to the cleavage of hydrolyzable linkages in the polymer chains. This process can produce oligomers or monomers that are further absorbable within the body. Oxidative reaction involves the release of reactive oxygen species and free radicals from inflammatory cells, causing depolymerization of the polymer.¹⁰

Fig. 12 Biodegradable properties. (a) Schematic illustrations showing the representative in vivo degradation mechanisms of bioresorbable polymers, including (i) dissolution, (ii) enzymolysis, (iii) hydrolysis, and (iv) oxidative degradation.³⁵² Reproduced from ref. 352 with permission from American Chemical Society, copyright 2020. (b) Design of a bioresorbable cardiac pacemaker, (i) polyurethane served as the encapsulating layer, Mo as the conductor material, and monocrystalline Si as the diode. (ii) In vitro degradation test showed that this device could be fully dissolved after 40 days in PBS solution. (iii) In vivo experiment demonstrated that it could provide efficient pacing even after implantation for 1 month.³¹² Reproduced from ref. 310 with permission from AAAS, copyright 2022.

In general, the degradation of bioresorbable synthetic polymers like polyvinyl alcohol (PVA), poly (lactic-co-glycolic acid) (PLGA), poly (L-lactic acid) (PLLA), and polycaprolactone (PCL) mainly occurs via hydrolysis, where hydrolyzable linkages including ester, carbonate, anhydride, and amide bonds in these polymers, are cleaved by water molecules. Then, these resultant degradation products can undergo absorption by the body. Simultaneously, synthetic polymers exhibit degradation processes characterized by either bulk erosion or surface erosion mechanisms. Bulk erosion involves the diffusion of water and/or enzymes into the material, leading to the gradual accumulation of degradation products within the polymer, which catalyzes continuous degradation and decreases the weight of the polymer. On the other hand, surface erosion entails the hydrolysis process occurring exclusively on the material surface, leading to a gradual and linear reduction in mass or volume over a period of time. These polymers are often utilized to form substrates, while metals such as magnesium (Mg), zinc (Zn), iron (Fe), tungsten (W), and molybdenum (Mo), which could be degradable by dissolution, hydrolysis, and oxidative reaction, can be used for circuits and electrodes. In addition, silicone (Si) can also serve as a viable material choice for bioresorbable devices, as it undergoes hydrolysis to silicic acid in vivo.³⁵⁴ An example of the complete structure and compositions of a transient stimulator could be seen in Fig. 12b. This bioresorbable pacemaker consisted of a bioresorbable polyurethane encapsulating layer, stretchable Mo electrodes, and a steroid-eluting patch.³¹² The device was thin, lightweight, and stretchable to minimize tissue irritation or damage, with tailored geometries that could be customized to fit the patient's anatomy. The bioresorbable Mo conductor enabled a functional lifetime of over one month under simulated physiological conditions. The steroid-eluting patch could release dexamethasone acetate (DMA) for several months to reduce local inflammation and fibrosis following cardiac pacing. This bioresorbable pacemaker was specifically engineered to ensure consistent and reliable cardiac pacing when attaching onto the mechanically dynamic heart surface, while concurrently minimizing any discernible deviations in output voltage during mechanical deformation. In vivo experiments demonstrated that this bioresorbable pacemaker could support temporary pacing therapies, and its full dissolution took more than 1 month.

To fulfill the demands of practical applications, the degradation rate of ES devices can be finely tailored by various strategies including the change in encapsulating layer thickness and modulation of physicochemical properties of bioresorbable polymers such as composition, molecular weight, and crystallinity.⁴⁹ For example, the increase in crystallinity, molecular weight, or hydrophobicity can reduce the degradation rates of polymers. In addition, it is feasible to adjust the composition of polymers such as the ratios of hydrophobic units to hydrophilic units to control the dissolution rates. For instance, elevating the glycolic acid (a hydrophilic unit) to lactic acid (a hydrophobic unit) ratio in PLGA can expedite the rate of mass loss in a physiological environment.³⁵⁵ In addition, some degradable building blocks (e.g., polymer chains or crosslinks) are often introduced into the design of hydrogel-based bioelectronic interfaces to improve their in vivo biodegradability.²¹⁶ Consequently, the degradation rate of hydrogel interfaces can be effectively modulated by manipulating the proportion of degradable building blocks to non-degradable counterparts.

4.6. Other key properties

Breathability assumes a paramount role for development of bioelectronic devices, particularly for on-skin bioelectronics, as the water vapor and sweat easily accumulate at the skin-device interfaces that may cause skin irritation and device failure during prolonged application.³⁵⁶ Fabrication of devices with porous structures has been demonstrated as a feasible approach to improve breathability. For example, Tian et al. fabricated a microperforated silicone layer by using poly(methyl methacrylate) microspheres as a sacrificial template, which could serve as a supporting substrate for Au mesh electrodes.³¹⁸ By tuning the density and size of the templated microspheres, the porosity of the silicone supporting substrate could be finely regulated. The water vapor transmission rate (WVTR) of such silicone layers surpassed 10 g m⁻² h⁻¹, exhibiting a substantially enhanced breathability as compared to the PDMS elastomer with an equivalent thickness.

Over recent years, the utilization of electrospun fibers for preparation of ultrathin breathable bioelectronics with a porous structure akin to fabric or textiles has garnered growing interest.^357–359 The fiber diameter and porosity of electrospun membranes can be readily adjusted by varying different process parameters such as concentration of precursor solution, voltage, collection distance, etc.³⁶⁰ For example, Ma et al. proposed a stretchable liquid metal-based poly(styrene-block-butadiene-block-styrene) electrospun membrane, realizing breathability and moisture permeability in the flexible devices. In another recent study, Tang et al. employed electrospun membranes with a sandwich structure as a self-powered ES platform, in which PCL/PLGA was the upper layer, PCL was the interlayer, and PPy was the under layer.³⁶¹ This ES patch demonstrated remarkable gas exchange capabilities, and its WVTR was significantly higher than that of solid PDMS films and commercial patches. Despite the excellent WVTR observed in these breathable bioelectronics under normal conditions, they may encounter limitations when worn by individuals who experience excessive sweating in hot and humid environments or engage in intense physical exercise, since these bioelectronics are unable to rapidly remove sweat and may result in subsequent functional failure. To tackle this issue, the integration of functional multilayer textiles and/or fabrics that support continuous sweat discharge presents a promising solution.³⁶² These textiles possess directional water transport properties, enabling the sweat transfer from the skin to the external surface of the bioelectronics, facilitating rapid evaporation.³⁶³ Simultaneously, the textile can possess waterproof properties to prevent the invasion of external water into the bioelectronics, thereby safeguarding the normal operation of the devices.³⁶² It should be emphasized that these breathable designs are highly suitable for electrodes, energy harvesters/storage, and/or sensors; however, direct integration of various conventional electronic components or encapsulation would impede the functionality of breathable channels. Recently, our group put forward a new highly integrated permeable wearable device by designing a 3D liquid diode configuration (Fig. 13a).³⁶⁴ The vertical liquid diode system was achieved by imparting a wettability gradient to a superhydrophobic fabric, while the horizontal liquid diode relied on the combined effects of a roughness gradient (wettability) and a structural gradient (curvature gradient). This 3D liquid diode enabled direct integration of stretchable circuitry, electrodes, electronic components, batteries, and RF antenna, while accomplishing effective and unidirectional sweat transportation, with a maximum flow rate that is 4000 times greater than the physiological sweat rate.

Fig. 13 Other key properties. (a) (i) A highly integrated breathable bioelectronic that (ii) could transport sweat via vertical and horizontal directions.³⁶⁴ Reproduced from ref. 364 with permission from Springer Nature, copyright 2024. (b) (i) A bioadhesive electronic interface with anisotropic swellability. This interface was prepared by (ii) a substrate-constrained drying process and could (iii) swell in the thickness direction without changing length and width dimensions. (iv) In vivo test showed its superior ability to stimulate rat's sciatic nerve stably.³⁶⁵ Reproduced from ref. 365 with permission from Springer Nature, copyright 2021. (c) (i) The mechanism of water-responsive supercontractile polymer films. (ii) The in vivo application of bioelectronic interfaces.¹⁰ Reproduced from ref. 10 with permission from Springer Nature, copyright 2023.

In addition to breathability, the controllable swellability of bioelectronic interfaces has also been instrumental for their performances, especially in the case of hydrogel-based bioelectronic interfaces.³⁶⁶ One consensus is that moderate swelling properties are a prerequisite for hydrogel interfaces; hydrogel-based interfaces should rapidly absorb interfacial water to enhance device–integration, while avoiding excessive swelling in highly humid environments (e.g., in vivo microenvironment) that may compromise their structural stability and mechanical performance.³²⁴ Deng et al. prepared a bioadhesive interface with anisotropic swellability by a substrate-constrained drying process (Fig. 13b).³⁶⁵ When coming into contact with a moist tissue surface, the bioadhesive interface could remove water rapidly to form a robust adhesion to tissue and simultaneously swell in the thickness direction without changing length and width dimensions. In vivo experiments further demonstrated stable and efficient ES of the sciatic nerve via this bioadhesive interface.

Stimuli-responsiveness represents a highly desirable and intriguing characteristic of many intelligent systems such as drug delivery platforms.³⁶⁷ In the specific context of ES systems, this capacity to undergo dynamic, stimulus-driven changes has been demonstrated to be an advantageous and beneficial performance attribute. The integration of stimuli-responsive mechanisms within ES platforms has accelerated the development of sophisticated, adaptive systems capable of tailoring their behaviors and properties in direct response to relevant environmental cues or triggers, thereby optimizing their functionality and efficacy for targeted therapeutic interventions. For example, conventional neurostimulation cuffs are flexible yet hard to achieve tight but soft contact with biological tissues, since they are soft without standardized shapes or sizes. To implement a generally applicable device-tissue interface that has great mechanical and electrical performance to help adapt and intimately wrap the arbitrary targeted tissue well, Yi et al. reported a water-responsive super-contractile polymer film. This dry film showed large and rapid contraction after activation by water and thus could be firstly wrapped on the tissue loosely, and then it would contract until it wrapped the tissue seamlessly (Fig. 13c).¹⁰ Taking inspiration from spider silk, the films were fabricated using a combination of poly(ethylene oxide) (PEO) and poly(ethylene glycol)-α-cyclodextrin inclusion complex. These films were specifically designed to possess aligned microporous hierarchical structures, enabling their integration with electronic components. The primary objective of this research was to address the challenge of developing stimuli-responsive films that are compatible with both soft tissues and electronic integration processes. The microstructure formation during mechanical elongation could illustrate the super-contraction mechanism in this film. The process involved the utilization of inclusion complex platelets that facilitated the permanent crosslinking of the isotropic semicrystalline domains of PEO through the formation of hydrogen bonds. During the uniaxial cold drawing process, the PEO domains underwent plastic deformation, resulting in the creation of aligned fibrillar connections and porous architectures. Simultaneously, the PEO crystallites and chains achieved orientation and temporary fixation due to the formation of newly formed PEO crystallites. Water played a critical role in this process by inducing PEO chain recoil and super-contraction, thereby disrupting the PEO crystallites. Subsequent to the contraction phase, the PEO crosslinked by the inclusion complex transitioned into an amorphous state with a high water content, transforming the films into hydrogel films. This intricate interplay of molecular interactions and structural changes led to the fascinating behavior of films when exposed to water. The stimulation performance demonstrated in the study showed promising results in terms of acute nerve stimulation, compound nerve action potentials and muscle action potentials recording, as well as in electroneurogram (ENG) recording.

Notably, there are many other performances (e.g., self-healing^368,369 or shape memory capabilities^10,53,370) that can be modified or armed with to further improve the application efficacy of ES electrodes. These representative studies will be discussed in the subsequent sections on specific biomedical applications.

5. ES based bioelectronics for tissue engineering

Tissue engineering is a multidisciplinary field that merges biology with engineering to develop functional substitutes for damaged tissues, striving to match the structure and function of healthy in vivo tissues.³⁷¹ Advancements in materials science, microelectronics, and biology have led to novel tissue regeneration breakthroughs, particularly through the use of electrical stimuli to guide cell growth and specialization.^{5,13,16,129,372} Electric fields play great roles in the functioning of all living organisms, which can influence the behaviors and functions of various cells, such as fibroblasts, osteoblasts, or cardiomyocytes, as discussed in Section 2.1.^373,374 This section will explore the critical roles and underlying mechanisms of ES in tissue regeneration, including skin healing, neuro regeneration, bone regeneration and drug delivery.^49,375–378

5.1. Skin healing

Skin, as the largest organ in the human body, serves a crucial function in safeguarding internal tissues and organs against infections and injuries. However, it also presents significant risks of damage and injury. Generally, the process of skin healing contains several phases of hemostasis, inflammation, proliferation, and remodeling.^379–381 As for most superficial skin wounds, they can heal spontaneously in a short period of time. However, in the context of chronic skin wounds, classified as wounds that do not progress through the healing stages within a 42-day period, the journey towards complete recovery is marked by a lengthy trajectory and intricate biological mechanisms.³⁸² Up to now, chronic wounds have been some of the predominant clinical conundrums on a global scale. Wound healing can be decelerated by various factors, including diabetes, vascular deficiencies, advanced age, and nutritional inadequacies. These issues have caused huge financial burdens on the healthcare systems around the world, substantially reducing patients’ life quality and increasing the risk of infection and mortality.³⁴⁷ As a result, various strategies have been developed to accelerate wound healing, especially for the chronic wounds, such as the anti-bacterial wound dressing^383–385 and ES therapy.^386,387

Decades ago, the theory of skin battery and the relationship between skin wounds and endogenous electrical signals were proposed by Barker³⁸⁸ and Ghadamali,³⁸⁹ respectively. They believed that the electric field intensity around the wounds was closely related to the wound size and edge distance. As the research moves along, it was found that the endogenous electric field on skin wounds was generated by the ion directional transport between polarized epithelial cells and such electric fields displayed a paramount role in directing cellular migration during the healing process of epithelial wounds.^381,390 ES can also enhance the production and remodeling of collagen, elastin, and other extracellular matrix components, which can improve the mechanical strength and elasticity of the skin.³⁹¹ It is necessary to develop exogenous ES generators for wound treatment for simulating or enhancing wound potential and thus accelerating wound healing.

According to the differences in current models, commonly used ES methods can be categorized as either the unidirectional current (or voltage) or bidirectional current (or voltage) stimulation method.³⁸¹ The unidirectional current has a constant polarity because of the unidirectional flow of charged particles, while the bidirectional current is characterized by reverse polarity (Fig. 14a). Unidirectional waveforms can generate the current in a single direction and thus simulate an endogenous electric field. However, the excessive unidirectional current over a long time may cause thermal effects and damage the skin. In contrast, the bidirectional current stimulation strategy contributes to the targeted healing effects, facilitating wound repair without imposing thermal stress. Consequently, it offers an innovative and efficient approach in wound management, providing therapeutic benefits while optimizing patient comfort.³⁸¹

Fig. 14 ES for skin healing. (a) ES waveforms of unidirectional current (charge-unbalanced) and bidirectional current.³⁸¹ Reproduced from ref. 381 with permission from John Wiley and Sons, copyright 2021. (b) The self-powered enzymatic microneedle patch for chronic wound healing.³⁷⁶ Reproduced from ref. 376 with permission from AAAS, copyright 2023. (c) The self-powered iTENG patch for accelerating wound healing.³⁹² Reproduced from ref. 392 with permission from Elsevier, copyright 2021. (d) Illustration showing the biological activities of the ePatch throughout the healing process when subjected to ES.³⁹³ Reproduced from ref. 393 with permission from Elsevier, copyright 2022. (e) Conceptual illustration of the PPA.³⁹⁴ Reproduced from ref. 394 with permission from John Wiley and Sons, copyright 2022. (f) Schematic and photographs of the NFC-based smart bandage.³⁷⁵ Reproduced from ref. 375 with permission from Springer Nature, copyright 2022. (g) Schematics of a wearable patch for infected chronic nonhealing wounds consisting of multiple sensors, drug delivery and ES electrodes.³⁷⁹ Reproduced from ref. 379 with permission from Creative Commons CC BY, copyright 2023.

Both unidirectional and bidirectional currents have similar effects on wound healing; the working modes are largely determined by the selection of power sources for these generators, especially for the wearable electronic wound dressing with emerging power selections. In recent years, the swift advancement of wearable electronics has provided optimized strategies for electronic wound dressing with light weight, excellent skin conformality and biocompatibility. However, the selection of suitable power sources is always an obstacle for further miniaturization and intergradation. Besides, traditional batteries like lithium batteries are bulky and large sized with risks of electrolyte leakage. As a result, emerging bioenergy has been widely used for the ES for accelerating wound healing. For example, enzymatic biofuel cells (EBFCs) are the emerging bioenergy devices that can transform chemical energy into electrical energy via the enzymatic catalysis process.^46,395,396 The biofuels (like glucose and lactate) are oxidized by enzymes and thus releasing electrons on the anode, while reduction reactions (like the reduction of oxygen) occur on the cathode by receiving electrons at the same. In this way, the complete electron pathway can be constructed between the anode and the cathode and thus generates directional current for accelerating tissue regeneration and wound healing. Zhang et al. designed and fabricated a self-powered enzymatic microneedle (MN) patch based on the principles of EBFCs.³⁷⁶ In this work, the patch was composed of anode and cathode MN arrays containing glucose oxidase (GOx) and horseradish peroxidase (HRP) encapsulated in zeolite imidazolate framework-8 (ZIF-8) nanoparticles as shown in Fig. 14b. Through the enzymatic cascade reaction within the MN patch, local hyperglycemia can be mitigated, along with generation of stable microcurrents. This promotes expedited healing and prevents diabetic wound scarring. Furthermore, the inclusion of ZIF-8 enhances the anti-bacterial capability and the immobilization performance of enzymes on the MN array.

Apart from the EBFCs, various self-powered ES systems utilizing nanogenerators have been fabricated. Generally, these generators are almost driven by the piezoelectric or triboelectric effect that enables conversion of mechanical energy to electrical energy. The development of TENGs was based on the principles of triboelectrification and electrostatic induction involving two disparate materials. The working modes of TENGs can be divided into four types: free-standing mode, in-plane sliding mode, single-electrode mode and vertical contact-separation mode.^397–400 On the other hand, the electric charge of PENGs was generated by the change of polarization due to the application of mechanical stress on piezoelectric materials.⁴⁰¹ For the application of wound healing, these nanogenerators can be designed with various structures and sizes using different materials for different body locations. As shown in Fig. 14c, Jeong et. al proposed a wearable ionic TENG (iTENG) patch for enhancing wound recovery, which consisted of a flexible, gel-based platform that captured biophysical energy and applied electrical potential to damaged tissue.³⁹² The iTENG generated temporary charge separation, creating an electrical potential difference between the biomedical patch and the affected wound area. At the cellular level, this electrokinetic phenomenon instigated the translocation of electrochemically charged ions via ion channels present in the epidermal layer, elicited by this ES. Furthermore, this perturbation of the transepithelial potential (TEP) stimulated an intrinsic electric field. This electric field served to modulate and guide the migratory trajectory of dermal cells—particularly keratinocytes, endothelial cells, and fibroblasts situated along the wound periphery—towards the locus of the wound. Although conductive hydrogels can be applied to the stimulation electrodes for wound healing, their ion conductivity is not sufficient for an excellent electrode and the water loss after long-time usage is another obstacle for clinical practices. Subsequently, Wang et al. proposed a flexible electrical patch (ePatch) featuring anti-bacterial and printable conductive hydrogel electrodes composed of AgNWs and methacrylated alginate (MAA) for accelerated wound healing (Fig. 14d).³⁹³ The formulation of the hydrogel is specifically tailored to facilitate printing on medical-grade patches, aimed at delivering personalized treatment for wounds. The ePatch notably enhances re-epithelization, stimulates neovascularization, modulates the immune response, and curbs infections within the wound's microenvironment. In vitro investigations have revealed a marked enhancement of cellular proliferation and migratory potential in response to ES. Apart from hydrogels, various materials have been reported for both wearable electronics and stimulated wound healing devices with great biocompatibility and skin adaptability. Among them, gallium-based liquid metals have been found to be excellent candidates for the development of soft and stretchable electronics due to the unique combination of fluid compliance and robust electrical conductivity.^402,403 The encapsulation and safe attachment of liquid metals on human skin have always been a primary focus in the realm of wearable electronics combined with liquid metals. As shown in Fig. 14e, Cheng et al. developed a straightforward method for creating wet-adhesive electronics, utilizing a metal–polymer conductor composed of a PEG-blended PDMS-based adhesive (PPA) that encases circuits made of gallium-based liquid metal alloys.³⁹⁴ The PPA is capable of bearing a larger deformation than PDMS and adhering conformally to the skin over two days with lower modulus. Besides, wet-adhesive electronics have excellent biosafety, which provides great choice for ES based electronics for wound healing. When pulsed ES was applied on the electrode attached to the wound area, the wound nearly healed in 10 days with higher density of collagen. Wound tissues revealed that treated areas developed scars by day 4, experienced fibroblast proliferation and vascular formation by day 7, and had mature epidermis and hair follicles by day 10.

Although these materials combined with ES sources can accelerate cell proliferation, migration and vascular formation, these wound dressings are passive and moderately effective. They cannot actively respond to variations in the wound environment or intelligently adjust the strength of therapy. The chemical composition of wound exudate changes substantially, reflecting wound conditions and healing stage. For example, both increased wound pH and temperature are the indicators of local wound infection.^404,405 Smart bandage technologies are suitably equipped to tackle these problems. They combine various types of sensors and stimulators, which enable constant monitoring and active wound management with the least amount of intervention from medical professionals.³⁷⁵ However, current smart ES devices are large with bulky batteries.⁴⁰⁶ As a result, a battery-free, wireless, closed-loop, smart bandage could continuously monitor skin impedance and temperature and deliver ES in response to the wound environment (Fig. 14f).³⁷⁵ Besides, a hydrogel interface was formulated, incorporating PEDOT:PSS and N-isopropylacrylamide (NIPAM). This development aimed to secure seamless skin integration and staunch electrical communication with the tissue. This interface exhibited a tunable temperature response capability, maintaining solid adhesion at ambient temperatures, but surrendering its adherence upon exposure to high temperatures. This phenomenon facilitates straightforward and gentle interface removal, preventing the infliction of additional harm to sensitive wounded tissue. Moreover, as shown in Fig. 14g, Sanl et al. demonstrated a more complex and stretchable wireless system integrating multiplexed sensors and treatments for chronic wounds.³⁷⁹ They combined various sensors, including temperature (T), pH, ammonium (NH₄⁺), glucose (Glu), lactate (Lac), and uric acid (UA) sensing electrodes, reference (Ref) and counter electrodes, and a pair of voltage-modulated electrodes for controlled drug release and ES on a stretchable electrode. For the data collection and control module, it was constructed on a FPCB to ensure its stability. The combination of multiple sensors and treatments was able to monitor wound conditions more precisely and accelerate wound healing.

5.2. Muscle regeneration

Muscle tissues are essential elements of an organism, making up more than 50% of body mass and playing key roles in generating force, facilitating body movement, and supporting the operation of internal organs.⁴⁰⁷ One key characteristic of muscle tissues is their ability to contract in response to electrical signals.⁴⁰⁷ Muscle tissues are categorized into three types: skeletal, responsible for body movement;^36,408 cardiac, specialized muscle tissue determining the heart contraction;^409,410 and smooth muscles, located in the vascular and digestive systems.^37,411 Different from skin wound healing, muscle regeneration is a process of repairing damaged muscle fibers. Healthy muscles have a strong capacity to regenerate from minor injuries through the activation, proliferation, and differentiation of muscle stem cells.⁴¹² However, severe injuries often lead to inadequate endogenous regeneration, resulting in fibrosis and scarring.⁴¹³ Currently, surgical reconstruction is the most prevalent therapy for muscle injuries; however, it is constrained by low survival rates and scarce donor tissue availability. Muscle regeneration, which constructs muscle tissues by integrating growth factors, cells and scaffolds, is a promising therapeutic strategy for treating muscle injuries.⁴¹² Scaffold design in this field aims to comprehensively mimic the natural extracellular matrix, incorporating structural, biochemical, and biophysical cues to foster appropriate cell growth.^386,414 In clinical settings, techniques like ES and physical exercise are proven to enhance muscle mass and strength.^415,416 Specifically, external ES effectively prevents muscle mass and strength loss due to atrophy, often seen after extended periods of immobility following injury or surgery.^417,418 Additionally, ES can affect various muscle cell properties, such as the proliferation rate and hypertrophy, by emulating the action of motoneurons in the physiological context.^419–421

Conductive biomaterials have shown promise in enhancing the proliferation and differentiation of cells that are sensitive to electrical stimuli.^44,407,422 These materials are excellent candidates for scaffolds in muscle regeneration because of their effective conductivity and ability to facilitate tissue formation.^407,414 Graphene and its derivatives are increasingly utilized as biomaterials owing to their unique electrical and mechanical properties. Jo et al. synthesized conductive graphene hydrogels from GO/polyacrylamide (GO/PAAm) composites, creating materials with muscle-like stiffness and significantly reduced electrochemical impedance.⁴²³ These hydrogels not only boosted the proliferation and differentiation of C2C12 myoblasts, but also enhanced myogenic gene expression under ES, making them potent scaffolds for skeletal muscle regeneration. Incorporating electroconductive hydrogels into traditional soft tissue engineering fulfills key requirements for accelerating tissue regeneration, including high conductivity, suitable mechanical properties, tissue adhesion, and biodegradability, thus facilitating precise treatments within healthcare systems. Compared to conventional hydrogels, injectable conductive hydrogels have seen significant advancements recently, because they can be effectively delivered to inaccessible sites for tissue regeneration.⁴²⁴ Jin et al. developed an electroactive, biocompatible and injectable hydrogel, based on phenylborate-mediated crosslinking and conductive Au nanoparticles, enabling immediate two-way electrical conduction in the muscular system (Fig. 15a).³⁶ This hydrogel has expedited tissue repair in rat models of severe muscle injury and has shown effectiveness in closed-loop, robot-assisted muscle rehabilitation. In addition to conductive materials, those capable of transforming mechanical energy into electrical energy are being explored to support the regeneration behavior of myoblasts in human tissues. Ribeiro et al. combined a piezoelectric polymer with magnetostrictive particles, demonstrating that both mechanical and ES from these composites enhance myoblast maturation,⁴²⁵ which established a promising method for skeletal muscle tissue regeneration by stimulating magnetoelectric cells.

Fig. 15 ES for muscle regeneration. (a) Injecting the hydrogel into the damaged muscle improved the tissue repair process (left). In vivo muscle-tissue conduction in the severe tissue defect model. Scale bar, 5 mm (top right). Images depict the sequential stages from the induction of VML injury to treatment with the IT-IC hydrogel and subsequent muscle tissue regeneration at four weeks. Scale bar, 1 cm (bottom right).³⁶ Reproduced from ref. 36 with permission from Springer Nature, copyright 2023. (b) The integration of ES and drug delivery in a biodegradable and implantable device for enhancing muscle regeneration.⁴⁹ Reproduced from ref. 49 with permission from American Chemical Society, copyright 2022. (c) Immunofluorescence staining of muscle myobundles with F-actin (green images) and SAA (red images).⁴⁰⁸ Reproduced from ref. 408 with permission from AAAS, copyright 2021. (d) Nerve-targeted neuromorphic ES for alleviating inflammation in tendon injuries.²⁵⁴ Reproduced from ref. 254 with permission from Springer Nature, copyright 2024. (e) Isolated cardiomyocytes cultured in alginate and nanowire composites demonstrate potential for synchronized activity and organized tissue formation across the scaffold, featuring cardiac cells (red), alginate pore walls (blue), and AuNWs (yellow), with signal evolution depicted through colors, contour lines, and arrows.⁴⁰⁹ Reproduced from ref. 409 with permission from Springer Nature, copyright 2011. (f) The application of the ES E-bandage for intestinal wound healing.³⁷ Reproduced from ref. 37 with permission from Springer Nature, copyright 2024.

In addition to materials that directly interface with muscle tissues to influence the electrical microenvironments around muscle injuries and transmit ES to target areas, bioelectronics with more functions are also important. Lee et al. demonstrated that EBFCs enhanced the migration, proliferation, and differentiation of muscle precursor cells.³⁸⁶ GOx generates the anodic current and bilirubin oxidase (BOD) produces cathodic currents in the EBFCs. ES from the EBFC significantly enhances cell migration, cell proliferation, and the differentiation of cells into myotubes, as evidenced by gene and protein expression levels.³⁸⁶ This EBFC system was later incorporated into a nanofiber scaffolding system, showcasing its potential for muscle repair and regeneration. Huang et al. developed an implantable, bioresorbable electronic medicine system using a MN device that can be activated wirelessly, combining ES with anti-inflammatory drug delivery (Fig. 15b).⁴⁹ In experiments on a rat skeletal muscle injury model, this system not only regulated cell behaviors and enhanced tissue regeneration through periodic ES but also controlled inflammation via sustained drug release. This integrated approach eliminates a necessary second surgery for removing the device, simplifying the treatment process. Skeletal muscle regeneration is significantly facilitated by both local and systemic inflammatory responses. Although inflammation after muscle injury typically aids in muscle tissue regeneration, uncontrolled inflammations in some situations like rheumatoid arthritis and dermatomyositis are linked with muscle deterioration and weakness. Chen et al. discovered that exercise-mimetic ES effectively counteracts the muscle weakness and atrophy triggered by chronic interferon-γ (IFN-γ) treatment (Fig. 15c).⁴⁰⁸ This ES not only induced myofiber hypertrophy but also mitigated IFN-γ-induced muscle degradation by downregulating the JAK (Janus kinase)/STAT1 (signal transducer and activator of transcription 1) signaling pathway, revealing a novel anti-inflammatory mechanism beneficial for muscle exercise and therapies. Furthermore, Bao et al. introduced a neuromorphic electro-stimulation technique that utilizes a circuit fabricated using an ultrathin semiconductor floating-gate memory for direct nerve stimulation (Fig. 15d).²⁵⁴ This method entails the encasement of the sympathetic chain within flexible electrodes that are programmed to emit bionic spikes. This approach led to a significant 73.5% reduction in the inflammatory cytokine IL-6, showcasing its potent anti-inflammatory effects in tendon regeneration.

The heart circulates blood through the whole body via the contraction of cardiac muscles. Unfortunately, cardiac muscle tissue has limited regeneration ability after injury, often leading to the replacement of lost cardiomyocytes with fibroblasts.^410,424,426 These fibroblasts form scar tissue, which can result in arrhythmias and heart remodeling. Dvir et al. explored an innovative approach that involves incorporating Au nanowires (AuNWs) into alginate scaffolds to electrically enhance communication between cardiac cells in engineered cardiac patches (Fig. 15e).⁴⁰⁹ This enhancement results in thicker tissues, better alignment, and the capability of synchronous contraction when electrically stimulated. Tissues cultivated on these matrices also display elevated levels of proteins crucial for muscle contraction and electrical coupling, potentially enhancing the therapeutic effectiveness of cardiac patches. Another significant focus within muscle tissue engineering is the regeneration of smooth muscles, which play a crucial role in controlling gastrointestinal motility.^15,23,411 Wu et al. developed an innovative ES device, termed the E-bandage, which was applied to intestinal wounds (Fig. 15f).³⁷ This E-bandage enhanced local production of the epidermal growth factor and promoted the proliferation of all intestinal tissue layers, including smooth muscles. As a result, this approach led to improved healing and reduced postoperative complications compared to conventional treatment with sutures.

5.3. Bone regeneration

Bone regeneration is the process of repairing or regenerating damaged or diseased bone tissues, which are the main structural and supportive components of the human body. As one of the few tissues in mammals, bone can generally regenerate on its own after fracturing. However, in some special cases, like congenital skeletal abnormalities and large volume bone infection, the self-regeneration process would be disrupted, which leads to delayed bone healing or nonunion.^427–429 Nonunion has caused huge burden on patients, resulting in continuing pain, restricted daily activities, and increased medical costs. As a result, it is necessary to find an effective strategy for bone regeneration and remodeling. In the field of bone tissue engineering, ES has been researched and effectively implemented in clinical settings to promote bone healing for many years. ES is able to induce osteogenesis, mineralization, and integration of the bone cells, such as osteoblasts, osteoclasts, or stem cells.⁴³⁰ ES can also promote the formation and growth of blood vessels and nerves, which can provide the nutrients and signals for the bone. ES devices for bone regeneration can be classified as bone growth stimulators, bone graft stimulators, and bone scaffold stimulators.^126,431

Similar to the in vitro stimulation platforms, according to the differences in power sources, the electrostimulators for bone regeneration could be divided into the invasive DC ESs and non-invasive stimulators based on the inductive coupling or capacitive coupling mechanisms (Fig. 16a).⁴²⁷ As for the DC stimulators, they are always surgically implantable and can deliver current to a region of interest and consist of a subcutaneously implantable current generator and electrode. The cathode location is the site of osteogenesis and thus is implanted in the target area,^427,432 while the anode should be placed near the soft tissue.^427,433 DC stimulators show a positive correlation between the effects and electrical energy parameters and are able to achieve a large amount of focal energy delivery.^434,435 However, there are potential safety risks during the invasive implantation process, such as the inflammatory response and infections. As a result, the noninvasive stimulators based on inductive coupling and capacitive coupling principles have been applied for bone ES. These stimulators do not need any invasive surgery for implantation. For inductive coupling stimulators, one or two coins are connected to the signal generator and applied to the skin. Based on Faraday's Law, a time-varying magnetic field is produced when an electrical current flows through a coil, which subsequently triggers the inception of an electric field within the biological tissue.⁴²⁷ The capacitive coupling stimulators need two conductive pads, also known as capacitive electrodes or plates, to be situated on the skin flanking the target area. The integrated power resource produces either a stable or fluctuating electrical field within the target tissue, situated between the capacitive plates.⁴²⁷ Correspondingly, noninvasive stimulation exhibits a response that correlates directly with the duration of the treatment, evidencing a dose-dependent effect.

Fig. 16 ES for bone regeneration. (a) Classification of ES based bone growth by the mechanism of operation and electrical energy delivery.⁴²⁷ Reproduced from ref. 427 with permission from IEEE, copyright 2018. (b) Schematics of the biodegradable ES device, and X-ray radiographs on the bone fracture area over time of the experimental group.³⁷⁷ Reproduced from ref. 377 with permission from CC BY 4.0 open access license, copyright 2021. (c) Schematics, size and implantation of the self-powered ES based on a TENG.⁴³⁶ Reproduced from ref. 436 with permission from Elsevier, copyright 2019. (d) Schematic illustration of a nano-conductive hydrogel combined with ES for inducing bone formation.⁴³⁷ Reproduced from ref. 437 with permission from Elsevier, copyright 2023. (e) Schematic of a smart ES scaffold, and H&E staining images of rabbit radial defects in various groups.⁴³⁸ Reproduced from ref. 438 with permission from Elsevier, copyright 2020.

For bone ES, the selection of power sources is still a matter that requires significant attention, especially for the implanted stimulators. In addition to the commonly used commercial batteries, there are also self-powered and even transient strategies for bone ES, especially for the implantable stimulators. Similar to the wound ES, up to now, most self-power stimulators for bone regeneration have been developed using EBFCs, TENGs and PENGs. As shown in Fig. 16b, a self-powered bioresorbable ES device for bone fracture healing was proposed.³⁷⁷ The implantable, biodegradable and flexible bone fracture ES device (FED) comprises a TENG to produce electricity, in conjunction with a duet of dressing electrodes designated for direct application of electrostimulation to the fracture site. The TENG component includes an island-bridge Mg bottom electrode with a micro-pyramid-structured PLGA layer on top, forming the bottom triboelectric layer. Another layer of PLGA coated with an island-bridge Mg electrode acts as the top triboelectric layer. The device is capable of adhering to uneven tissue surfaces and generates biphasic electric pulses triggered by body movements. In vivo, the FED remained stable and unchanged during the first 8 weeks, followed by rapid degradation and resorption within 14 weeks. The X-ray radiographs showed that the FED is capable of accelerating fracture healing, enhancing bone callus development, facilitating bone marrow clearance, and statistically improving mineral density and flexural stress of the fractured bone. Moreover, another self-powered, implantable electrical stimulator has been presented, designed to enhance osteoblast proliferation and differentiation by utilizing a nanostructured polytetrafluoroethylene film in conjunction with a micro-structured aluminum film (Fig. 16c).⁴³⁶ For this TENG, the application of an external force instigated contact and friction between two layers, culminating in the formation of triboelectric charges of opposing polarities on their surfaces. Upon the withdrawal of the external force, electrons within the attached Au induction electrode and the micro-structured aluminum film executed a reciprocal motion via the external circuit, equilibrating the electrical potential difference induced by the tribocharges. This process yielded an electrical signal that is modulated in accordance with the external force applied. The results showed that ES provided by the device significantly promoted osteoblast attachment, proliferation, and differentiation. Intracellular calcium levels increased following ES, which played a critical role in regulating functions of osteoblasts. Beyond applying different materials for the nanogenerator, developing potential materials for ES electrodes is also a point of concern for bone regeneration. As an illustration, an innovative conductive hydrogel, encompassing PEDOT:PSS-magnesium titanate-methacrylated alginate infused with calcium phosphate (CPM@MA), was synthesized with the intention of facilitating electro-inspired bone tissue regeneration (Fig. 16d).⁴³⁷ Exhibiting superior electroactivity, biocompatibility, and osteoinductivity, the CPM@MA hydrogel notably augmented cellular functionality. This was accomplished through the upregulation of endogenous transforming growth factor-beta1 (TGF-β1) and the activation of the associated TGF-β/Smad2 signaling pathway. The hydrogel, when combined with ES, promoted intracellular calcium enrichment and significantly enhanced bone defect regeneration in vivo. Furthermore, as shown in Fig. 16e, Cui et al. designed an electroactive composite scaffold by integrating a triblock copolymer of poly(L-lactic acid)block-aniline pentamer-block-poly(L-lactic acid) (PLA-AP) with PLGA/hydroxyapatite (PLGA/HA) for bone repair that incorporates the controlled release of human bone morphogenetic protein-4 (hBMP-4) upon ES.⁴³⁸ The ES can be used not only for osteogenesis differentiation, but also for drug delivery. By incorporating growth factors or genes into tissue engineering scaffolds, it was possible to create a platform that can provide sustained release of bioactive molecules, respond to specific stimuli for on-demand release, and improve the overall efficacy of the treatment. Additionally, the combination of scaffold and drug delivery systems could address challenges such as complex purification processes, high costs, and potential side effects associated with direct application of bioactive agents. Upon comparing the hematoxylin and eosin (H&E) staining results, it was found that the PLGA/HA/PLA-AP/phBMP-4 composite exhibited the highest calcium staining intensity. Moreover, electrically stimulated scaffolds exhibited more robust staining intensities when compared with scaffolds that did not receive any electrical stimulation. These results suggested that the electroactive material could show osteoinduction under ES and the scaffolds containing bioactive molecules exhibited excellent osteogenic properties even without ES.

5.4. Nerve regeneration

Nerve injuries can cause extensive motor, sensory, and autonomic dysfunctions, severely impacting daily and professional activities.^439–441 The nervous system possesses regenerative capabilities. However, achieving complete functional recovery is infrequent and depends on various factors, including injury severity, timing of repair, proximity to the target organ, patient's age, and axonal guidance.^7,442,443 Complex injuries demand intricate repair approaches and prompt surgery mitigates degenerative risks, but axonal growth rates and distance to the target organ can prolong recovery, with younger patients generally faring better in regeneration.^444,445 The limited growth rate of axons leads to an extended recovery period when there is a significant separation between the site of injury and the intended target. This prolonged period increases the risk of permanent muscular atrophy or fibrotic changes.⁴⁴⁶ Integral to functional restoration is the axons’ capacity to navigate and reinnervate the precise targets, and any aberration in this process can lead to unsatisfactory recovery outcomes.⁴⁴⁷ Ultimately, the precise reinnervation of targets is crucial for restoration, and deviations in this process can greatly affect the overall success of functional recovery after nerve injury.

Optimal nerve recovery is often hindered by factors such as chronic axotomy, chronic denervation, and slow axonal growth at surgical sites. Research has shown that low-frequency (20 Hz) ES applied after surgical repair can significantly enhance nerve regeneration, evidenced by accelerated reinnervation and improved motor axonal regeneration in both animal and clinical studies.⁴⁴⁸ The underlying biological mechanisms that facilitate the regenerative effects of ES include the promotion of axonal growth through cyclic adenosine monophosphate (cAMP), enhancement by neurotrophic factors, and the involvement of Schwann cells (Fig. 17a).⁷ The application of ES of 20 Hz for a duration of one hour on injured peripheral nerves has been found to increase neuronal cAMP levels and expedite the enhanced expression of neurotrophic factors and their receptors in both neurons and Schwann cells. This activity resulted in elevated expression of cytoskeletal and growth-associated proteins, along with the release of neurotrophic factors from Schwann cells, particularly nerve growth factor, which is crucial for stimulating axon outgrowth from the proximal nerve stump of injured neurons. Experiments have demonstrated the critical roles of brain-derived neurotrophic factor and neurotrophin-4/5 in enhancing the effectiveness of ES for nerve regeneration post-injury and surgical repair using transgenic mouse models.

Fig. 17 ES for nerve regeneration. (a) Neurons and Schwann cells transition into a growth state, where motoneurons display chromatolytic changes such as the nucleus shifting to an eccentric position, indicative of heightened gene expression of various regeneration-associated genes (RAGs) in both neurons and Schwann cells.⁷ Reproduced from ref. 7 with permission from John Wiley and Sons, copyright 2015. (b) The fully implantable neural ES (FI-NES) system operates by delivering targeted electrical impulses directly to the site of long-segment peripheral nerve injury in vivo, promoting accelerated nerve regeneration and functional reconstruction through enhanced neural activity and growth stimulation⁴⁴⁹ Reproduced from ref. 449 with permission from John Wiley and Sons, copyright 2021. (c) A schematic depiction shows a tubular Zn–O₂ battery encircling a nerve, serving as an in vivo power source to facilitate nerve regeneration.⁴⁵⁰ Reproduced from ref. 450 with permission from John Wiley and Sons, copyright 2023. (d) The patch is designed for direct stimulation of the targeted nerve bundle, as shown in a photograph featuring the neural stimulation patch enclosed in plastic tubing with a diameter of 3.2 mm.⁴⁵¹ Reproduced from ref. 451 with permission from Creative Commons CC BY license, copyright 2023. (e) The implantable PHBV/PLLA/KNN film nanogenerator, used in conjunction with ultrasound, delivers in vivo ES to effectively enhance peripheral nerve repair.⁴⁵² Reproduced from ref. 452 with permission from Elsevier, copyright 2022. (f) A schematic representation of a device designed for sciatic nerve regeneration shows its composition, which includes porous PCL, PLLA-PTMC, a Mg–FeMn galvanic cell, and electrospun PCL fibers.⁴⁵³ Reproduced from ref. 453 with permission from AAAS, copyright 2020. (g) A biodegradable device design is illustrated, featuring a wireless receiver, serpentine stretchable extension electrodes, and a stimulation cuff with terminal exposed electrodes.³⁷⁸ Reproduced from ref. 378 with permission from Creative Commons CC BY license, copyright 2020.

In clinical practice, a variety of advanced strategies, including precise nerve repair surgeries and pioneering bioengineering approaches like nerve guidance conduits, are employed to enhance peripheral nerve regeneration and provide structural support for nerve growth while reducing complications and concentrating growth-promoting factors.^454–457 The administration of direct ES to the site of neural injury has emerged as a compelling adjunctive modality to bolster neurologic recovery. This therapeutic intervention harnesses electrical currents to augment the reparative processes of compromised peripheral nerves. It can substantially elevate the expression of critical neurotrophic factors, notably nerve growth factors, which are pivotal in ensuring neuronal survival and stimulating axonal sprouting.^458,459 Concurrently, ES has been found to enhance the synthesis of cAMP within neuronal tissues. This quintessential intracellular messenger is instrumental in activating an array of signaling cascades that expedite the neural recuperative sequence and axonal regrowth.^460,461 The regenerative benefits conferred by ES have been demonstrated across a spectrum of nerve injury paradigms, encompassing compressive traumas,^462,463 complete transections,^464,465 and lesions spanning extensive distances.⁴⁶⁶

NGCs are used in combination with ES to treat neural injury, providing physical support and directional guidance while enhancing nerve growth and functional recovery through ES. This combination therapy enhances both the precision and effectiveness of nerve regeneration, making it one of the most promising methods to improve functional recovery following peripheral nerve injury.^467,468 In order to realize NGCs combined with ES, the electrical conductivity of the catheter is an important property. Researchers have explored a repertoire of conductive materials, including CNTs,^66,469 graphene,^470,471 and conductive polymers such as PPy,^{102,472–474} PANi,^475,476 and PEDOT,^477,478 to fabricate NGCs that facilitate accelerated peripheral nerve regeneration.

Cumbersome external percutaneous wiring is an important issue for implantable ES devices because it can lead to infection risks, local discomfort, activity restrictions, and maintenance difficulties. In response to these constraints, research has pivoted towards the development of wireless implantable ES devices, employing various methods of energy supply to circumvent these issues. Generating ES based on the piezoelectric effect of the device itself is a novel method.^{452,479–481} Wu et al. reported a self-powered piezoelectric polymer scaffold embedded with zinc oxide (ZnO) nanogenerators for addressing traumatic neural injury.⁴⁷⁹ Upon ultrasound stimulation, they observed that Schwann cells cultured on this scaffold exhibited enhanced process elongation and upregulated expression of proliferation markers, neural proteins, and angiogenic factors. Implantation of the device improved neurological deficits in mouse models under the stimulation of running. Expanding upon self-powered solutions, devices based on triboelectric nanogenerators⁴⁸² or hybrids of triboelectric and piezoelectric nanogenerators⁴⁴⁹ have been engineered. Jin et al. proposed a friction/piezoelectric hybrid nanogenerator capable of generating electrical signals that were physiologically self-regulated. Upon subcutaneous implantation into the chest, the device yielded therapeutic outcomes for long-segment peripheral nerve injuries (15 mm defect) that were on par with those from autologous nerve grafts (Fig. 17b).⁴⁴⁹ Moreover, tubular zinc-oxygen (Zn–O₂) batteries were explored as an in situ power source for ES during nerve repair. These batteries address the compatibility issues of neural scaffolds with rigid lithium batteries⁴⁸³ and the suboptimal electrochemical performance of glucose fuel cells,⁴⁸⁴ achieving 231.4 mWh cm⁻³ volumetric energy density within a minimal volume of 0.86 mm³ (Fig. 17c).⁴⁵⁰ They can be seamlessly incorporated into nerve conduits to deliver ES directly to the site of injury. An alternative strategy involves wirelessly powering battery-free NGCs from outside the body using electromagnetic induction^451,485 or RF^378,486 energy. Jensen et al. reported a wireless electromagnetic nerve stimulation patch that promoted nerve regeneration through stimulation generated by electromagnetic induction (Fig. 17d).⁴⁵¹ This device functioned by capturing energy through a biocompatible inductive coil, made by sputter-coating Au onto a structure of PCL fibers. It can operate at emitter voltages between 2 V and 6 V and frequencies ranging from 10 kHz to 100 kHz, producing therapeutic voltages between 10 mV and 60 mV. Furthermore, Koo et al. introduced a platform that incorporated an RF power harvester for electrical peripheral nerve stimulation.⁴⁸⁶ This wireless stimulator contained an energy harvester composed of a Mg coil inductor, RF diode with a Si nanofilm active layer, Mg electrode, and Mg/SiO₂/Mg capacitor. The Mg electrode embedded within a PLGA sleeve facilitated connection to the peripheral nerves in rodent models, successfully enhancing nerve regeneration and functional recovery.

The necessity of secondary surgical interventions for the extraction of implantable devices represents a significant limitation, often resulting in patient discomfort, pain, and potential complications.⁴⁸⁷ Consequently, the development of bioabsorbable ES devices has become a critical avenue of research. Bioresorbable ES constructs, employing piezoelectric nanogenerators⁴⁵² and galvanic cells,⁴⁸⁸ have been investigated for their potential to obviate the need for device retrieval. Wu et al. developed piezoelectric and biodegradable materials, including potassium sodium niobate (KNN) nanowires, PLLA, and poly(3-hydroxybutyrate-co-3-hydroxypentanoate) (PHBV), coupled with poly(lactic acid) or PCL films as encapsulation layers, and Mg electrodes and Mo wires (Fig. 17e).⁴⁵²In vivo studies demonstrated that such an ES catheter, activated by ultrasound, promoted rat sciatic nerve regeneration. Similarly, Wang et al. explored the integration of biodegradable galvanic cells with nerve guidance conduits to provide both structural reinforcement and ES for the regeneration of peripheral nerves.⁴⁸⁸ The device comprised thin-film Mg and iron-manganese (FeMn) alloy electrodes, with conduits fashioned from porous PCL and copolymers of PLLA and poly(propylene carbonate) (PLLA-PTMC), yielding a flexible and highly stretchable interface (Fig. 17f).⁴⁵³In vivo experimentation in rat models with sciatic nerve transection injuries featuring a 10 mm gap showcased the device's efficacy in accelerating nerve regeneration and functional recovery. An emerging strategy involves the utilization of battery-free electrical stimulators, which are powered by external RF energy.^378,486 Such devices are fabricated using bioresorbable materials, including Mg, Mo, and PLGA. Choi et al. reported a bioresorbable polyurethane as both a substrate for the ES device and a barrier against biofluids (Fig. 17g).³⁷⁸ When used in conjunction with additional biodegradable inorganic materials, including Si, SiO₂, Mg, and Mo, this device proved effective in facilitating neuromuscular regeneration, as evidenced by its capacity to mitigate muscle atrophy following denervation in a rat peripheral nerve injury model. Besides invasive electrodes, non-invasive techniques also can transmit electrical current into the injury parts for accelerating the regeneration. Wu et al. introduced a concept involving the use of capacitive coupling hydrogels to deliver in situ ES to nerves injured in spinal trauma. This approach is aimed at enhancing remyelination, hastening axonal regeneration, and supporting the differentiation of endogenous neural stem cells.⁴⁷

5.5. Drug delivery

The field of drug delivery has experienced significant advancements, with the emergence of electric fields as a promising technique with diverse therapeutic applications.⁴⁸⁹ Such applications encompass actuating drug depots for targeted in vivo delivery, facilitating the transport of drug molecules through electroporation and iontophoresis techniques, and extending to the direct treatment of cancerous tumors.^490,491 Drug delivery can be enhanced by ES, which can increase the permeability and transport of the cell membrane, by creating temporary pores or channels. ES can also control the release and uptake of the molecules, by modulating the electrical potential or charge of the cell membrane.^492,493 According to the size of the delivered molecules, electric fields can improve effectiveness through electroporation and iontophoresis for transdermal delivery, respectively.

5.5.1. Iontophoresis. Iontophoresis enables the transportation of small molecules across the skin layer known as the stratum corneum via electrophoresis and electroosmosis at low-voltage (typically <10 V) (Fig. 18a).⁴⁹⁰ As low current and voltage is applied, the iontophoresis process barely alters the permeability of skin. This technique has been widely used in various realms, like drug delivery and sweat extraction. For example, as shown in Fig. 18b, a stretchable Mg–air battery devised by Huang et al. utilizes a dual-ion conducting hydrogel (SDICH), showing intelligent biomedical applications in drug delivery and wound healing.⁴⁹⁴ This battery incorporated advanced mechanical design and fabrication methodologies, using an SDICH consisting of choline chloride (ChCl) and sodium nitrate (NaNO₃). It displayed stable electrical outputs irrespective of 40% stretching. This attribute permitted functionality under a variety of mechanical deformations without degradation of performance. Subsequently, a self-adaptive wound dressing containing antibiotics was developed, derived from the stretchable battery and SDICH. Therein, a thermal resistor and the drug loaded SDICH as the loads were serially connected to the battery. When infection occurred on the wound, the skin temperature increased while the resistance of the thermal resistor would decrease. As a result, the voltage on SDICH would increase and thus more antibiotics were released by iontophoresis effects for accelerating wound healing. In vitro tests on agar and in vivo tests on mouse models have been conducted to characterize this system, indicating the effectiveness of the intelligent self-adaptive wound dressing. In addition to the proposed biocompatible soft batteries, some self-powered technologies have also been developed for iontophoresis. Similar to wound healing and bone regeneration, the common self-powered strategies are also based on EBFCs, TENGs and PENGs. As shown in Fig. 18c, an organic transdermal iontophoresis patch with a built-in biofuel cell was proposed.⁴⁹⁵ The patch was composed of carbon fabrics modified with enzymes, a resistor based on conducting polymers, and hydrogel films that encompass the biofuel along with the deliverable chemicals. Additionally, it featured a PDMS frame, enclosed by an oxygen-permeable medical tape. The patch generated a transdermal ionic current that created osmotic flow from the anode to the cathode, thereby delivering chemicals into the skin. The performance of the patch was tested on pigskin and found to be stable for 6 hours. The patch was also tested on a man's wrist and showed flexibility and stability during wrist motion. The transdermal ionic current was found to enhance the penetration of skincare compounds into the skin. Besides, Wu et al. reported a self-powered wearable iontophoretic transdermal drug delivery system that is driven and controlled by the energy harvested from biomechanical motions.⁴⁹⁶ As shown in Fig. 18d, the assembled system comprised a wearable TENG serving a dual role as a motion sensor and as an energy harvester, paired with a hydrogel-based soft patch supplemented with side-by-side electrodes for a noninvasive iontophoretic transdermal drug delivery system. The functionality of this system was validated via proof-of-concept trials using pigskin with dyes as surrogate drugs. The proposed system broadened the applicability of TENGs in biomedical realms, offering an economical alternative for noninvasive, electronically enhanced transdermal drug delivery, with integrated sensing and treatment modalities.

Fig. 18 ES for drug delivery by iontophoresis. (a) The working principle of iontophoresis involves using a low electrical current to transport charged molecules on the skin's surface.⁴⁹⁰ Reproduced from ref. 490 with permission from Springer Nature, copyright 2021. (b) The concept and operational principle of the self-adapting wound dressing system, along with the corresponding fluorescence images of agar samples from control and experimental groups.⁴⁹⁴ Reproduced from ref. 494 with permission from Creative Commons CC BY license, copyright 2022. (c) Photograph of the organic transdermal iontophoresis patch based on EBFCs mounted on a human arm.⁴⁹⁵ Reproduced from ref. 495 with permission from John Wiley and Sons, copyright 2014. (d) Optical images of a wearable transdermal drug delivery system achieved using a TENG.⁴⁹⁶ Reproduced from ref. 496 with permission from John Wiley and Sons, copyright 2019. (e) The photographs provide a structural overview of the integration of the biobattery, consisting of four cells connected in series, with the porous microneedle.⁴⁹⁷ Reproduced from ref. 497 with permission from Creative Commons CC BY license, copyright 2021. (f) Optical image of a wireless SOP and PLGA micro needle array encapsulated by an electrically triggerable layer.⁴⁹⁸ Reproduced from ref. 498 with permission from Creative Commons CC BY license, copyright 2024. (g) Illustration of the MN platform for real-time and in situ diabetes monitoring and treatment.⁴⁹⁹ Reproduced from ref. 499 with permission from Creative Commons CC BY license, copyright 2021. (h) Illustration of the close-loop wound monitoring and drug delivery electrode array.⁵⁰⁰ Reproduced from ref. 500 with permission from John Wiley and Sons, copyright 2021.

Subsequently, the same group applied a PENG based controllable transdermal drug delivery system for treating psoriasis.⁵⁰¹ In this study, to counter the low drug penetration resulting from skin thickening, the authors constructed smart, electrically-responsive MNs to foster drug penetration, integrating a conductive substance, PPy. The self-powered, manageable transdermal drug delivery system (scTDDS), grounded on a PENG, could modulate drug release, transforming mechanical energy into electrical energy. This illustrates effective drug dissemination and therapeutic impacts for conditions resembling psoriasis. In addition, Kusama et al. also developed a solid polymer-based ion-conductive porous microneedle (PMN) array for improving transdermal drug delivery through iontophoresis.⁴⁹⁷ As shown in Fig. 18e, the PMN array was modified with the charged hydrogel of poly-glycidyl methacrylate (PGMA) with excellent mechanical stability and biocompatibility. Notably, the study also demonstrated the enhanced efficiency of transdermal molecular penetration and extraction using the MN array, as well as the potential for powering the system with EBFCs. The results showed that the PMN was able to lower the transdermal resistance, generate a larger EOF and transport larger molecules through interconnected micropores, making it a promising option for advanced transdermal iontophoresis. To further improve the intelligence and spatiotemporal resolution of MN based transdermal drug delivery, a spatiotemporal on-demand patch (SOP) was established (Fig. 18f).⁴⁹⁸ Drug-loaded MNs are integrated with SOP, featuring biocompatible metallic membranes that can transmit an electrical field to the MNs, triggering drug release. This setup allows for precise control of the drug release amount to target locations. The SOP was also customizable and scalable for various pharmaceutical needs with multi-domain large-scaled patches for high-resolution drug delivery.

Furthermore, integrated closed-loop feedback systems, which combine the strengths of sensor technologies and iontophoresis-based drug delivery techniques, have emerged as a promising strategy for advancing healthcare solutions and intelligent personalized medicine devices.^502–504 These sophisticated systems can realize intelligent monitoring and regulation of patient status, creating a sustainable self-adjusting feedback loop that ensures timed therapeutic intervention and optimal drug delivery. The efficient cooperation between the sensor technology and iontophoresis drug delivery system allows for a streamlined, patient-specific treatment regimen. By continuously monitoring, analyzing, and reacting to the patient's status, the integrated closed-loop system creates a personalized therapy map, attaining a favorable balance between disease management and patient comfort.^502,504 As demonstrated in Fig. 18g, Li et al. proposed a fully integrated closed-loop system for smart blood glucose monitoring and diabetes treatment.⁴⁹⁹ They constructed this system based on mesoporous microneedles (MMNs). The system consists of three main components: a reverse iontophoretic glucose sensor using MMNs, a flexible printed circuit board (FPCB) for integration and control, and an iontophoretic insulin delivery device using MMNs. The MMNs enabled painless penetration of the skin, facilitating glucose extraction and insulin delivery. The system has been shown to accurately monitor glucose fluctuations and deliver insulin to regulate hyperglycemia in a diabetic rat model. In addition, a digital wound dressing was proposed combining uric acid, pH and temperature sensors with the iontophoresis drug delivery strategy (Fig. 18h).⁵⁰⁰ The smart wound dressing was able to simultaneously detect the temperature, pH, and uric acid levels at the wound site, which could be used to assess the wound condition. The drug delivery electrode in the dressing was coated with the antibiotic cefazolin and could release the drug by applying a voltage (∼0.5 V) to the electrode, achieving on-demand infection treatment. Especially, a near-field communication (NFC) technology was applied to this integrated system for avoiding battery use and maintaining small size.

5.5.2. Electroporation. Electroporation has been one of the most commonly used strategies for non-viral drug or gene delivery by applying high-voltage electrical pulses, which can create temporary pores or channels in the cell membrane. Electroporation can be used for various purposes, such as gene therapy, cell reprogramming, or immunotherapy. Electroporation-based drug delivery can be optimized by adjusting the parameters of the electrical pulses, such as amplitude, duration, frequency, and number. Bulk electroporation (BEP) is an accessible and cost-efficient technology that operates on a user-friendly setup. In this procedure, cells and their corresponding cargos are initially placed in a dielectric chamber, resulting in the formation of a transmembrane potential across the lipid bilayer (Fig. 19a).⁵⁰⁵ When the potential reaches a critical value (by applying large pulsed voltage), the lipid molecules in the membrane would undergo reformation to generate minor openings on the cell membrane, which is capable of accelerating the translocation of cargos into the internal area of cells via processes of diffusion and endocytosis.^22,505 Although the BEP can process millions of cells, this strategy requires a substantial voltage (typically >100 V), leading to inconsistent poration across individual cells. To tackle these issues, Dong et al. developed a microwell array-based electro-delivery nanochip. This innovative nanoplatform was utilized to analyze gene characteristics and drug resistance across millions of individual cells, as shown in Fig. 19b.⁵⁰⁶ The platform employed a Domino-probe as a means of distinguishing intracellular target RNAs, with the subsequent amplification of accompanying fluorescence signals. The incorporation of an on-chip addressable microwell-nanopore array proved instrumental in the electro-delivery process of the Domino-probe, as well as facilitating in situ monitoring of cellular conduct. The on-chip allocated microwell-nanopore assembly served a pivotal function in the nanoplatform designed for individual living cell analysis. This arrangement facilitated the secured and homogenous intracellular dissemination of the Domino-probe, a DNA probe used for identifying gene mutations. Through this microwell structure, in situ cell cultures, intracellular drug delivery, and spatiotemporally regulated cellular behaviors could be investigated. The nanoporous membrane, with openings measuring 800 nm in diameter, restricted the electric field to the cells placed accordingly, resulting in perforation of the cellular membrane and electrophoretic delivery of Domino-probes within the cells. When benchmarked against conventional BEP, this electro-delivery mechanism was competent in delivering the Domino-probe with superior efficiency, consistency and safety employing reduced voltages.

Fig. 19 ES for electroporation. (a) Schematic of a bulk electroporation system. The gap between the electrodes exceeds the size of a single cell by multiple orders of magnitude.⁵⁰⁵ Reproduced from ref. 505 with permission from Royal Society of Chemistry, copyright 2016. (b) The schematic illustrates a microwell array-based nanochip for detecting target RNA in single living cells and provides a cross-sectional representation of a nanoplatform that electrically delivers the Domino-probe into cells and pinpoints their locations.⁵⁰⁶ Reproduced from ref. 506 with permission from American Chemical Society, copyright 2021. (c) Working principle of electroporation for transdermal drug delivery.⁴⁹⁰ Reproduced from ref. 490 with permission from Springer Nature, copyright 2021. (d) Schematic illustrations depicting the application of a SEFM, including its structure, and the transdermal delivery of nicotinamide using the SEFM on the inner forearms of subjects.⁵⁰⁷ Reproduced from ref. 507 with permission from John Wiley and Sons, copyright 2023. (e) The design and optical images feature an electroporator equipped with a piezoelectric pulse generator and MNs, along with the components of the ePatch.⁵⁰⁸ Reproduced from ref. 508 with permission from Creative Commons Attribution License 4.0 (CC BY), copyright 2021. (f) Photos display the flexible MN array electrode (MNAE) chip, showcasing: 1, the MNAE attached to a glass substrate; 2, the flexibility of the MNAE; and 3, a close-up view of silicon MNs coated with Au.⁵⁰⁹ Reproduced from ref. 509 with permission from Royal Society of Chemistry, copyright 2021.

As shown in Fig. 19c, high-voltage pulses were applied on the skin with pulse width on the order of microseconds to milliseconds, and a dielectric breakdown at the stratum corneum subsequently punctured the skin surface, thereby enhancing its permeability.⁴⁹⁰ Immediately following this, the voltage was reduced to prevent any potential stimulation of the subjacent tissue. However, the disruption of the lipid bilayer structures in the skin is not always reversible. Therefore, researchers have been focusing on the development of more effective electroporation methods using low voltages in recent years. Xu et al. introduced a stretchable electronic facial mask (SEFM) for skin electroporation with great skin adaptability, reusability, portability, and low cost (Fig. 19d).⁵⁰⁷ In this work, a conductive ink exhibiting skin adhesion, negligible resistance, and notable stretchability was developed, comprising components such as carbon black, graphene, silicone oil, kerosene, and a Pt catalyst. Notably, the silicone oil additive in the conductive ink played a critical role in augmenting the adhesive bond between the ink and silicone substrate. This culminated in efficient bonding and crosslinking, facilitating the ink electrodes to withstand deformation, in tandem with the silicone, under stretching conditions without succumbing to fracturing. In this system, the electrical pulse was tunable including the amplitude and frequency, and the in vivo test on rats and human arms showed good biosafety.

To further break through the skin barrier for drug delivery, MNs have been combined with the electroporation technology, which can largely improve delivery efficiency of large molecules and decrease the skin disruption caused by high voltage. As shown in Fig. 19e, a portable electroporator combined with MNs (ePatch) was proposed for transdermal delivery of a DNA vaccine against SARS-CoV-2.⁵⁰⁸ The system combined a thumb-operated piezoelectric pulser adapted from a standard household stove lighter. The ePatch system could elicit robust antibody responses to SARS-CoV-2 in mice and allowed for at least a 10-fold reduction in dosage compared to traditional injection techniques. Notably, the ePatch system utilized microsecond pulses from a piezoelectric pulser, which minimized tissue heating compared to millisecond pulses and thus reduced damage to the skin. Besides, Wei et al. fabricated a wearable and flexible MN array combining with electroporation for in vivo DNA and siRNA delivery.⁵⁰⁹ As shown in Fig. 19f, the construction of the flexible MN array electrode (MNAE) chip was realized on a soft substrate, thereby facilitating its effective adaptation to varying tissue surfaces, which in turn assured even electroporation efficiency across the entire functional region of the MNAE under low voltage conditions.⁵⁰⁹ Displaying notable transfection efficiencies, ranging between 65% and 80%, for the green fluorescent protein (GFP) plasmid transfection, the electroporated cells retained optimal viability. Moreover, the effective penetration depth derived from low-voltage (30 V) electroporation could extend up to as much as 700 μm, demonstrating adequacy for a vast majority of nucleic biotherapeutic applications. The optimum voltage for efficient DNA transfection was around 40 V, beyond which an increase in voltage leads to a decrease in mean fluorescence intensity, possibly due to tissue and cell damage affecting DNA expression. This study addressed the challenges of in vivo nucleic acid delivery using MN arrays using low voltage and provided a novel strategy for efficient and safe delivery of nucleic acids to tissues or organs.

6. ES based bioelectronics for neuromodulation

Neuromodulation via ES is a promising medical technique that activates, suppresses, or modifies nerve activity with electrical impulses.^8,510 This approach offers hope to patients with a range of neurological and psychiatric conditions.^6,8,511 Through targeted ES of nerves or specific brain regions, neuromodulation can influence neural circuits, conducive to symptom alleviation and functional restoration.^512–514 Brain stimulation encompasses various methods that either activate or inhibit brain function through both direct and indirect means.^515–518 Brain ES methods range from non-invasive options such as transcranial current stimulation,⁵¹⁹ to invasive modalities like deep brain stimulation (DBS)^520,521 and intracranial ES (IES),⁵²² which involves the surgical placement of electrodes within the brain. The strategic implementation of ES presents valuable opportunities for mitigating symptoms associated with neurological and psychiatric disorders that are unresponsive to conventional therapies. Meanwhile, this approach offers insights into the intricate mechanisms of the brain and holds promise for improving patient outcomes.^24,513,523 Spinal cord stimulation (SCS) involves administering electrical currents to the spinal cord or associated nerves.^433,524 This innovative technique aims to obstruct neural signals before they reach the brain or to reestablish neural pathways between the brain and the peripheral nervous system.^50,525 SCS primarily serves as a therapeutic modality for the management of chronic pain⁵²⁶ and the rehabilitation of mobility,⁵²⁷ while modulating the function of specific organs.⁵²⁸ The vagus nerve, in particular, originates in the brainstem and traverses down to the abdomen, interfacing with the heart, lungs, and digestive system.^529,530 Peripheral neuromodulation targets the nerves located outside the brain and spinal cord.^11,531 Both the vagus nerve and peripheral nerves share a similar elongated and slender anatomical structure. Various forms of peripheral nerve stimulation (PNS) and vagus nerve stimulation (VNS) employ electrical impulses to regulate the activity of nerves associated with organ function or muscle control, expanding the potential for therapeutic intervention.^530,532,533 Each neuromodulation technique leverages the fundamental principles of ES to achieve specific therapeutic effects, showcasing the versatility and potential of this technology to either enhance functionality or elevate the well-being of individuals dealing with diverse health issues.^534,535

6.1. Mechanisms of ES for neuromodulation

The nervous system comprises numerous neurons, which are the essential units for transmitting excitatory and inhibitory signals.⁵³⁶ Neurons are composed of several parts: the cell body (soma), dendrites, initial segments, axonal segments (including nodes of Ranvier and internodes), and terminal partitions that connect to other neurons or tissues via synapses.^14,537 In 1939, Hodgkin and Huxley inserted an electrode into the giant axon of a squid, revealing that electrical signals, known as action potentials, conduct neural activity.⁵² Their groundbreaking work in 1952 on membrane voltage significantly advanced our understanding of how action potentials propagate along nerve fibers.²⁰⁴ For instance, when a mechanoreceptor such as the Pacinian corpuscle is stimulated by external pressure, it generates a voltage at the axon hillock (Fig. 20a).^4,538 When the generator voltage exceeds a certain value, it initiates the action potential that propagates along the axonal membrane, manifesting as an electrical impulse in the sensory neuron.⁴

Fig. 20 Mechanism of ES on a neuron. (a) Schematic illustration of efferent neurons (left) and afferent (right) neurons and the flows of excitation.⁴ (b) Representation and equivalent circuit modelling of ES of myelinated nerves.⁴ (c) Illustration of current exiting the membrane where the current flux is most concentrated.⁴ (d) Computed reactions of a neuron stimulated from within the soma (blue) and from an external microelectrode (orange).¹⁸ Reproduced from ref. 18 with permission from Elsevier, copyright 1999. (e) Response of the SENN model of a 20-μm myelinated fiber.⁴ (f) Strength–duration response of both current and charge to rectangular stimuli with varying phase duration and polarity.⁴ Reproduced from ref. 4 with permission from Artech House, copyright 2010.

ES can directly impact and modulate neural activities, either exciting or inhibiting the neuron.^539,540 Likewise, when externally applied voltage surpasses the threshold level (rheobase), it triggers the opening of voltage-sensitive sodium channels on the membrane.^43,536 This influx of Na⁺ increases the membrane voltage to the level of an action potential, which then propagates without further electrical input. Nerve fibers can be classified into myelinated and unmyelinated types (A- and C-fibers, respectively), based on whether the axon is covered by myelin sheaths.⁵³⁸ Compared to the neuron's soma, which has a higher membrane capacitance and is less excitable, the axon-especially at the exposed nodes—conducts electrical impulses more readily due to its lower capacitance and the insulating properties of the myelin sheath.

To understand how to effectively excite a neuron, researchers have developed mathematical models to quantitatively describe neuronal electrical characteristics. One model is the spatially extended nonlinear node (SENN) model, based on Donald McNeal's equations.⁵⁴¹ This model considers the resistance of the axoplasmic fluid (R_a), the capacitance (C_m), and resistance (R_m) of the axon membrane at each node, and the potential source (E_r) due to ion concentration differences. The model allows for the calculation of voltage propagation within the membrane, considering the external voltage distribution and the electrical properties of each node (Fig. 20b).⁴ It suggests that focusing the electrical field locally can lead to more efficient neuronal excitation, particularly near small electrodes where high spatial gradients occur. Besides this mode, even in uniformly distributed electrical field, the local depolarization could be triggered at ends (sensory receptors or motor-neuron end plates) or bends (where the axon is bent to a specific angle) (Fig. 20c).⁴

From F. Rattay's computational results, the electrical signal propagation along the neuron could be intuitively visualized, where the electrical impulse (duration: 0.1 ms) is generated from both inside the soma and external microelectrode (Fig. 20d, upper: inside, lower: outside).¹⁸ Internal negative currents result in hyperpolarization and inhibited activity, while positive currents can trigger action potentials. Extracellular stimulation, however, requires higher current intensities and can cause both hyperpolarized and depolarized regions within an axon, with intense reactions typically near the ends. Notably, strong negative extracellular currents exceeding 2.5 mA can block neural signals. The SENN model also demonstrates that rheobase-dependent action potential propagation can occur when the current reaches a certain threshold, leading to significant depolarization and subsequent action potential propagation along the axon. The solid orange curves in Fig. 20e showed the transmembrane voltage responses (V_n) at the same node of a stimulated fiber with a diameter of 20 μm. The stimulator electrode is point-shaped, placed 2 mm away, and the pulse width is 0.1 ms.⁴V_n = 0 refers to the resting potential of −70 mV. Thus, a positive response means depolarization and a negative response means hyperpolarization. While stimulating with pulse using 80% amplitude of threshold, V_n rises a little but drops down to 0 after the pulse quickly. When the current is increased to threshold (I_T) or above I_T, V_n exhibits a spike of above 100 mV after the stimulation pulse is finished. The dashed blue lines show the V_n at the next three nodes, where a spike propagating velocity of ∼43 m s⁻¹ along the axon could be deduced.

The concept of threshold rheobase is intricately linked to the duration of stimulation, a relationship known as the strength–duration curve (SD curve).⁵⁴² Using the SENN model to simulate responses to a spherical electrode positioned 2 cm from the neuron, it was observed that monophasic cathodic pulses typically require a lower threshold compared to biphasic pulses. Additionally, as the pulse duration extends, the threshold amplitude, quantified by peak current intensity and phase charge, shows markedly divergent trends, especially evident in monophasic stimulations (Fig. 20f).⁴ With higher pulse duration, the current threshold is much lowered, but the charge consumed is much increased. Thus, there is a balance between these two parameters and the optimal pulse duration that consumes lease electrical energy is called the SD time constant (τ_e).⁴³ After simplification and deduction of formulas that fit the SD curves, τ_e can be determined by simply using Q₀/I₀, where Q₀ refers to the minimum charge (fitted value) and I₀ the minimum current. This phenomenon reveals the reason why neurons respond stronger to stimulations with a certain frequency/phase duration, while too long or too short durations won’t work.

6.2. Electrical brain stimulation

Electrical brain stimulation represents a cutting-edge approach for both therapeutic intervention and neurological research.^3,512,517 By administering precise electrical currents to designated brain regions, this technique can modulate neuronal activity, potentially restoring or augmenting neural functions disrupted by conditions.⁵⁴³ ES of the brain can modulate sensory processing, modulating sensation by targeting structures such as the sensory thalamus or sensory cortex.⁵⁴⁴

6.2.1. Deep brain stimulation. DBS exerts its influence on neurotransmitter release and synaptic plasticity through a complex mechanism.^520,545,546 A crucial feature is that by directly suppressing or stimulating neural activities, high-frequency DBS (≥100 pulse per s, 0.1 ms per pulse) can effectively override various pathological oscillations, including Parkinson's disease,⁵⁴⁷ tremor,⁵⁴⁸ epilepsy,⁵¹² depression,^545,549etc. This mechanism explains the immediate impact of DBS in managing certain movement disorders but doesn’t fully explain the long-term adaptive changes seen in patients with dystonia and psychiatric conditions like depression. In animal studies, DBS has been shown to influence neurotransmitter release and synaptic plasticity, initiating calcium waves and gliotransmitter release upon stimulation, which affects synaptic plasticity, promotes arteriole dilation, and increases regional blood flow (Fig. 21a).⁵⁴⁶

Fig. 21 Deep brain stimulation (DBS). (a) Mechanisms of DBS. Stimulation induces the release of neurotransmitters, initiating calcium waves and the subsequent release of gliotransmitters, with DBS altering local field potentials in the subthalamic nucleus.⁵⁴⁶ Reproduced from ref. 546 with permission from Springer Nature, copyright 2019. (b) A typical DBS electrode (Medtronic model 3387) is placed in the STN and ZI (green), close to the thalamus (blue), substantia nigra (orange and yellow), and striatum/pallidum (red).⁵⁵⁰ Reproduced from ref. 550 with permission from The American Physiological Society, copyright 2016. (c) The time frames for the effects of deep brain stimulation.⁵²¹ Reproduced from ref. 521 with permission from Springer Nature, copyright 2017. (d) A three-dimensional visualization of activity, captured via MEG, in a human brain with an implanted DBS electrode (left),² and an illustration showing the CC lead and HD lead with a detailed view of the 40 contacts.⁵⁵¹ Reproduced from ref. 2 with permission from Springer Nature, copyright 1969. Reproduced from ref. 551 with permission from IOP Publishing, copyright 2004. (e) An image of highly flexible, meandering stimulation electrodes in water and a diagram explaining that theoretically, the absence of glial scarring minimizes the distance between the stimulation site and neurons, allowing for lower current requirements for more precise neural activation.⁵⁵² Reproduced from ref. 552 with permission from Elsevier, copyright 2023. (f) An image displaying typical mesh electronics, featuring recording electrodes (red marks) and unipolar stimulation electrodes (black marks). Scale bar, 200 μm. A schematic depicts a mouse implanted with mesh electronics connected via conductive ink to a pliable flat cable (red arrow).⁵⁵³ Reproduced from ref. 553 with permission from Springer Nature, copyright 2016. (g) Illustration of the GF bipolar microelectrode, which is compatible with the DBS-fMRI study.⁵⁵⁴ Reproduced from ref. 554 with permission from Springer Nature, copyright 2020.

The outcomes of stimulating various brain areas can differ markedly. For example, targeting the area that includes the dorsal subthalamic nucleus (STN) and zona incerta (ZI) has been associated with reducing Parkinson's disease symptoms like rigidity, bradykinesia, and tremor. Conversely, stimulating nearby structures can lead to side effects such as muscle contractions, diminished executive function, and mood disorders. (Fig. 21b).^550,555 In clinical settings, the STN and the internal segment of the globus pallidus (GPi) are primarily targeted by DBS to mitigate symptoms of Parkinson's disease.⁵⁴⁷ Both targets have been associated with similar improvements in motor symptoms, dyskinesia, and overall quality of life.^556,557 However, STN stimulation may be linked to heightened cognitive and mood complications, albeit with a greater potential for medication reduction. On the other hand, GPi stimulation tends to better manage axial symptoms, such as speech and swallowing difficulties. These distinctions highlight the importance of comprehensive patient evaluation by a multidisciplinary team to tailor surgical interventions for optimal results. The administration of brain ES may occasionally result in muscle contractions or twitches, particularly noticeable when a generalized seizure is purposefully induced during treatment.⁵⁵⁸

In epilepsy models, ES of specific brain regions like the mammillothalamic tract (MMT) or the anterior nucleus of the thalamus (ANT) showed anti-epileptic effects, highlighting the role of DBS in modulating neural circuits involved in seizure control.⁵¹² Essential tremor patients receive DBS in the ventral intermediate nucleus of the thalamus to reduce involuntary shaking.⁵⁵⁹ In cases of dystonia, stimulating the GPi internus helps ease muscle contractions and improve movement control.⁵⁵⁸ DBS also shows effectiveness in severe obsessive-compulsive disorder by targeting neural circuits involved in the disorder, leading to symptom reduction.⁵⁶⁰ For epilepsy, DBS may decrease seizure frequency and intensity by modulating brain activity through electrode placement in the anterior nucleus of the thalamus or other specific areas.⁵⁶¹

Moreover, DBS has been adopted to regulate mood-related behaviors by targeting specific brain regions that control emotions. Stimulating specific brain regions like the nucleus accumbens and lateral habenula can selectively enhance motivation and reduce anxiety, while targeting the ventromedial prefrontal cortex increases pleasure and reduces despair, highlighting the need for tailored DBS approaches in treating complex conditions like depression.^513,545 Villard et al. emphasized the significant roles of evoking pleasant sensations by electrically stimulating the dorsal anterior insula and amygdala, particularly in the right hemisphere.^562,563 It's crucial to note that DBS treatment necessitates careful patient selection, thorough assessment, and precise adjustment of stimulation settings by a skilled healthcare team.

Different mechanisms of DBS exhibit varying time frames in terms of symptom improvement across different neurological conditions (Fig. 21c).⁵²¹ Tremor and rigidity respond almost immediately to DBS, often improving within minutes, which suggests a rapid modulation of neural circuits. In contrast, bradykinesia shows a slower response, requiring several hours to improve, likely due to processes like neuronal reorganization or neuroplasticity. On a much longer scale, dystonia and mood changes in depression can take months to resolve, implying complex long-term changes in neural circuits and possibly structural brain reorganization. These diverse response times across symptoms indicate that DBS operates through various therapeutic mechanisms, not just through direct inhibition or excitation of targeted axons. This variability highlights the necessity for tailored DBS therapies that consider the specific temporal dynamics and mechanisms associated with each condition.

The left image in Fig. 21d shows a brain tissue cross-section with an electrode implant, underscoring the targeted region of this intervention for Parkinson's disease.² DBS relies on the precise implantation of electrodes into particular brain regions to treat various neurological disorders. The right image in Fig. 21d depicts the intricate electrode design that highlights the customized structure in DBS device manufacturing.⁵⁵¹ Leads with multiple independently controllable contacts allow for programmed targeting within the brain.⁵⁶⁴ Compared with conventional cylindrical leads, the advanced high-density leads are engineered to provide more nuanced control of the electric field for stimulation, aimed at improving selectivity and effectiveness, particularly in treating Parkinson's disease.⁵⁵¹ The electrodes’ capability to selectively target neural populations while minimizing activation of the surrounding fibers is of significant clinical interest and central to the study, as it enables more tailored stimulation, reducing side effects and enhancing therapeutic outcomes. Displacement of a DBS lead can cause unintended electrical currents to stimulate adjacent nerve fibers, potentially inducing adverse effects such as movement disorders,^565,566 speech impediments,⁵⁶⁷ cognitive difficulties,⁵⁶⁸ and mood fluctuations.⁵⁴⁹ To mitigate these risks, careful control over the spread of the electric current is essential, along with the ability to adjust the DBS lead's placement as necessary, without resorting to physical relocation.⁵⁶⁹

The long-term efficacy of neural probes is often hindered by mechanical mismatches and immune responses, which can lead to complications such as movement at the probe-tissue interface, glial scarring, and neuron loss.^570,571 These complications often degrade the probes’ recording and stimulation capabilities, sometimes in a matter of days to weeks. In attempts to rectify recording instabilities, probes may be repositioned, which can exacerbate tissue damage and disrupt the targeting of specific neurons, thereby constraining their long-term usability.⁵⁷² Ultra-flexible electrodes (Fig. 21e) promise better integration with the nervous system, diminishing mechanical mismatches.⁵⁵² These electrodes provide high-resolution, stable neural modulation over extended periods with minimal current, showing resilience and the ability to precisely activate neurons and evoke behavioral responses with currents as low as 2 mA. Designed for minimal tissue damage and reduced side effects, ultra-flexible electrodes offer a reliable, precise solution for neural repair and circuitry modulation.⁵⁷³ To combat chronic instability, it is suggested that the size and mechanical characteristics of probes should more closely resemble those of the surrounding neural tissue.^574,575 Innovations in mesh electronics,⁵⁷⁶ with physical properties akin to neural tissue (Fig. 21f), have demonstrated the ability to overcome the drawbacks of traditional probes, enabling stable, consistent in vivo recordings and stimulation of individual neurons over periods of up to at least 8 months.^1,553

Combining deep brain stimulation with imaging methods, such as functional magnetic resonance imaging (fMRI), can be challenging due to conventional metallic electrodes causing magnetic interference and artifacts, which obscure functional and structural brain mapping near the electrodes.^577,578 Choosing appropriate conductive materials to fabricate ES electrodes can introduce new functions or compatibility.^24,579 In Fig. 21g, researchers have utilized graphene fiber (GF) microelectrodes to enable comprehensive and impartial mapping of activation patterns during DBS, in conjunction with fMRI. These GF electrodes demonstrated a high capacity for charge injection, stability under stimulation, and compatibility with MRI.⁵⁵⁴ When used for DBS aimed at the subthalamic nucleus, GF electrodes have significantly enhanced motor symptoms in Parkinson's disease models and did not produce fMRI artifacts, facilitating detailed brain mapping during stimulation. The results suggest that DBS with GF electrodes affects motor and non-motor pathways, showing considerable potential as a method for translational research into the workings and outcomes of DBS therapy.

Although DBS is widely used in clinical treatments, it is often invasive, and the detailed mechanisms are not fully understood, making it almost inevitably risky and potentially accompanied by complications and side effects. These may include cord or wire malfunctions, but more concerning are the unknown and unpredictable side effects due to stimulation, which may activate or inhibit off-target brain regions. Brain regions are closely situated and sometimes difficult to delineate, leading to unintended consequences. Reported side effects in movement disorder treatments, such as Parkinson's disease targeting the STN, include hypomania,⁵⁸⁰ depression (sometimes with psychosis),⁵⁸¹ confusion,⁵⁸² hypersexuality,⁵⁸³ suicidal tendencies,⁵⁸⁴ hallucinations apathy,⁵⁸⁵ and uncontrollable crying.⁵⁸⁶ For Gilles de la Tourette syndrome treatments targeting the thalamus or GPi internus, side effects include fatigue and decreased libido.^566,587 In affective disorder treatments, side effects include changes in musical preferences,⁵⁸⁸ forgetfulness,⁵⁸⁹ and language disorders⁵⁹⁰ in obsessive-compulsive disorder treatments targeting the nucleus accumbens, and suicidal tendencies in depression treatments targeting the subcallosal cingulate gyrus.

Despite the high occurrence of side effects in DBS, it has been found that these incidents are related to stimulation directions and tend to occur when DBS exceeds certain thresholds, particularly in STN stimulation for Parkinson's disease treatments.⁵⁹¹ Choosing directional DBS in the optimal direction (posteromedial stimulation in half of the cases) can effectively increase the side effect threshold, enlarging the therapeutic window and mitigating side effects. Side effects caused by implanted electrode materials, such as fibrous capsule formation due to foreign body reactions, can be alleviated by incorporating functional interfacial materials. Wu et al. recently developed an adhesive anti-fibrotic hydrogel that can serve as the interface between implantable electrodes and tissue.⁵⁹² By forming a conformal interface, the infiltration of inflammatory cells into the interface is largely blocked, preventing fibrous capsule formation. These findings suggest that side effects associated with ES can potentially be fully avoided through careful selection of the most optimal stimulation parameters and electrode configurations.

6.2.2. Other electrical brain stimulation techniques. Deep brain techniques often use penetrating electrodes for brain stimulation, which can inflict brain damage. Transcranial ES (TES) applies electrical pulses to the scalp to tune neural activity in the brain.⁵⁹³ The configuration of TES generally involves positioning electrodes on the scalp, which then deliver electrical currents to designated areas of the brain (Fig. 22a).^519,594 Subcutaneous and epidural ES apply electrical current directly to the outer layer of neural tissue of the brain, which can be done for various purposes, such as mapping cortical functions, treating neurological disorders, and enhancing neuroplasticity for rehabilitation purposes.⁵⁹⁵ TES encompasses a range of techniques aimed at modulating neuronal firing patterns and network activity, which can lead to diverse physiological and behavioral outcomes,⁵⁹⁶ including transcranial direct current stimulation (tDCS) and transcranial alternating current stimulation (tACS).^597,598 These methods are continually refined to improve the transmission of currents to the brain by enhancing current density and focality as well as minimizing peripheral side effects. TES also aims to target specific brain rhythms through advanced closed-loop stimulation methods. Furthermore, it is crucial to address hardware and signal processing challenges during TES to prevent amplification saturation and effectively separate stimulation artifacts from the recorded physiological signals.

Fig. 22 Transcranial ES. (a) Diagram depicting the three configuration for transcutaneous (scalp), subcutaneous (skull) and epidural (dura) stimulation.⁵¹⁹ (b) Neuronal compartment orientation influences TES-induced excitability in four idealized neurons.⁵⁹⁴ (c) Five hypothesized mechanisms modulate online spiking in response to varying TES magnitudes.⁵⁹⁴ (d) The diversion of applied current by skin, soft tissue, and skull.⁵⁹⁴ (e) The distribution and intensity of the electric field within the brain during subcutaneous or transcutaneous stimulation.⁵¹⁹ Reproduced from ref. 519 with permission from Creative Commons CC BY license, copyright 2018. Reproduced from ref. 594 with permission from Creative Commons CC BY license, copyright 2018.

tDCS entails applying a steady, low-intensity electrical current to the scalp. This technique effectively modulates neuronal excitability by polarizing the underlying brain tissue. This method is favored for its simplicity, safety, and affordability, leading to its widespread adoption in both research and clinical environments (Fig. 22b).⁵⁹⁴ tDCS modulates neuronal activity by depolarizing neurons under the anode and hyperpolarizing them under the cathode, which can either enhance or suppress activity in targeted brain regions. This modulation influences changes in synaptic efficacy and neurotransmitter release, potentially altering levels of brain-derived neurotrophic factor and affecting synaptic plasticity and motor learning. Furthermore, tDCS influences gamma-aminobutyric acid (GABA) neurotransmission, enhancing its application in areas such as motor skill development, stroke rehabilitation, and the management of various neurological and psychiatric conditions.

On the other hand, tACS applies alternating current at specific frequencies to entrain brain oscillations and modulate neural activity, aiming to synchronize with existing brain rhythms to enhance cognitive functions and influence behavior. The immediate effects of tACS include changes in neuronal firing rates, increased neurotransmitter release, and ectopically induced spikes, which can alter the brain's native oscillatory patterns for cognitive and therapeutic purposes. Additionally, tACS influences network activity through mechanisms such as stochastic and rhythm resonance, temporal biasing of spikes, and network entrainment, which can either cooperate with or compete against endogenous brain activity (Fig. 22c).⁵⁹⁴ The impact of tACS on neural activity can be measured using methods such as spike and local field potential measurements, which provide insights into the acute effects of tACS on neural activity and network patterns. tACS has been investigated for its potential in modulating different frequency bands, such as gamma, beta, alpha, theta, and slow oscillations, to target specific cognitive processes and brain networks.

The key difference between tDCS and tACS lies in their mechanisms of action: tDCS modulates neuronal excitability through polarization, while tACS entrains endogenous brain oscillations through frequency-specific stimulation. In terms of application scenarios, tDCS is commonly used for modulating overall brain excitability and has been explored in various cognitive and motor tasks, as well as in clinical interventions for conditions like depression and chronic pain. On the other hand, tACS is more focused on entraining specific brain oscillations to target cognitive functions and neural networks associated with different frequency bands.

One critical consideration in TES is the shunting effect. The shunting effect during TES refers to the tendency of the current to divert away from the brain, following the path of least resistance, rather than efficiently reaching the brain surface, due to the skull's significant higher resistance (Fig. 22d).⁵⁹⁴ This diversion results in only a fraction of the delivered current actually reaching the brain surface, significantly influencing the effectiveness of the stimulation. There is also some controversy about the effectiveness of currently used tACS protocols on local neuronal networks. This controversy stems from challenges such as the externally induced electrical potential differences, which often hinder simultaneous measurement of brain signals. It's also hard to completely understand how different parts of the head affect current spread, and thus still can’t translate results obtained from experimental models and animal studies to humans. Furthermore, the translation of experimental findings to human applications is complicated by uncertainties in determining the current levels needed to affect neuronal patterns in the brain. These challenges highlight the need for further research and refinement in the application of tACS for targeted TES effects.

To tackle this issue, M. Vöröslakos and colleagues developed an intersectional short pulse (ISP) technique that enables the direct administration of increased current intensities into the brain, while reducing the side effects of scalp stimulation.⁵¹⁹ ISP is implemented by applying spatio-temporally rotating short pulses of electrical currents through multiple electrode pairs. This method demonstrated spatial specificity by activating single neurons in rats and directly affecting brain networks in healthy subjects. By exploiting the time-integrating properties of neuronal membranes, the ISP method focused on building up transmembrane charge at intersecting electric fields, thereby reducing adverse effects on other areas. Additionally, the ISP technique facilitated the measurement of the physiological impacts of scalp stimulation in human subjects without overwhelming the recording amplifiers. The study further confirmed the effectiveness of this method in rodents (Fig. 22e),⁵¹⁹ demonstrating that the amplitude of spontaneous local field potentials was influenced by the induced fields. Overall, the ISP method aimed to achieve higher intensity currents in the brain to affect neuronal circuits directly and instantaneously, addressing the limitations of conventional tACS approaches.

Compared with TES, cortical ES (CES) inserts electrodes directly on/into the cortical tissue. CES requires surgical exposure of the brain and is often used during neurosurgical procedures for brain mapping or as part of a therapeutic regimen in certain neurological disorders. Thin-film electrodes provide a flexible, conformal design that can be implanted in contoured or irregular brain areas, enabling more precise targeting and stimulation.⁹ As depicted in Fig. 23a, ultrathin epidural electronics were seamlessly integrated on the surface of a rat brain.⁵⁹⁹ This integration allows for comprehensive sampling of brain signals for diagnostic purposes, as well as the precise administration of therapeutic electrical pulses. This represents a significant advancement in the invasive management of epilepsy.⁵⁹⁹ Conventional thin-film electrodes are a less invasive option but suffer from inadequate resolution due to their distance from axonal processes. Fig. 23b illustrates that PEDOT:PSS electrodes, with a high density (40-μm pitch) and high capacitance (greater than 1 nF), achieve single-neuron resolution.⁶⁰⁰ These electrodes, fabricated on a parylene thin-film substrate, can be subdurally implanted and conform to the brain's surface for extended use.

Fig. 23 Electronics for cortical ES. (a) Schematic depiction of a graphene-based seizure sensor and epilepsy treatment stimulation sensors.⁵⁹⁹ Reproduced from ref. 599 with permission from John Wiley and Sons, copyright 2018. (b) A 3D array is applied directly onto the brain's surface.⁶⁰⁰ Reproduced from ref. 600 with permission from AAAS, copyright 2022. (c) Illustrative of hand's response to intracortical stimulation in different areas.⁶⁰¹ Reproduced from ref. 601 with permission from AAAS, copyright 2016. (d) The diagram illustrates a Stentrode, showing a self-expanding nitinol frame with Pt electrodes and a cross-sectional angiogram detailing the positioning of a delivery sheath, a delivery catheter, and electrodes on the scaffold, marked by a 10 mm scale bar.⁶⁰² Reproduced from ref. 602 with permission from Springer Nature, copyright 2018. (e) Design of a transcutaneous RF power and data transmission system for brain stimulation.⁶⁰³ Reproduced from ref. 603 with permission from Springer Nature, copyright 2021. (f) A bidirectional brain-computer interface that captures neural signals to generate a control signal for an output device and delivers sensory feedback by stimulating the brain in response to sensor data from the output device.⁵⁹⁵ Reproduced from ref. 595 with permission from Elsevier, copyright 2016.

Brain ES can induce sensory experiences, which are linked to the specific brain regions stimulated and the configuration of the ES.⁶⁰⁴ Cortical stimulation within the somatosensory cortex has successfully evoked tactile sensations.⁶⁰⁵ These perceptions, akin to pressure on the hand, align with the brain's somatotopic map and can be elicited using low stimulation amplitudes (Fig. 23c).⁶⁰¹ This capability shows great promise for the development of sensory neuroprosthetics, potentially restoring the sensory feedback necessary for complex hand movements and object manipulation.⁶⁰⁶ A study involving 20 adults found that electrical brain stimulation on different positions caused a variety of different sensations.⁶⁰⁷ The Monash Vision Group, which is at the forefront of creating a bionic vision system, implanted electrode arrays directly into the visual cortex designed to return sight to those with complete blindness.⁶⁰⁸ Each array features 43 active electrodes, alongside an electronic mechanism that is wirelessly powered to decode incoming signals and stimulate the brain with biphasic pulses.⁶⁰⁹ Brain ES may sometimes result in mild discomfort or pain at the stimulation site or nearby areas, likely due to nerve fiber activation or neuronal responses.⁵⁶⁶ Healthcare providers carefully calibrate the stimulation settings to alleviate any patient discomfort. Additionally, brain ES can trigger emotional reactions, ranging from euphoria to mood swings or emotional instability, though these effects usually dissipate once the stimulation ceases. It is crucial to recognize that sensations during brain ES are carefully overseen by medical professionals.

Traditional invasive electrode arrays for brain stimulation involve risky and invasive surgeries that may cause complications like inflammation and device failure. These methods offer high-resolution and quality signals, but they come with the cost of opening the skull. To reduce these risks, researchers have developed endovascular electrodes that can be used temporarily in less invasive procedures.^610,611 A stent-mounted electrode array has been shown to safely provide high-quality neural signals over extended periods in animal models.⁶¹² Opie et al. developed a stent electrode array for placement in the blood vessels over the motor cortex of sheep using a non-surgical approach, intending to demonstrate the feasibility of endovascular neural stimulation (Fig. 23d).^602,613 The study explored the stent-like electrode's ability to target specific cortical areas and examined how the device's position and electrode orientation affected its effectiveness. However, its effectiveness in stimulating precise brain regions from within blood vessels remains uncertain. Each type of electrode offers unique benefits and is carefully chosen based on the patient's specific requirements and the intended target within the brain.⁵³⁴ The selection process for an electrode is guided by several critical factors, including the neurological condition being addressed, the anticipated therapeutic goals, and the patient's unique anatomical structure. Neurosurgeons and healthcare experts evaluate these elements meticulously to identify the most appropriate electrode for each individual patient receiving deep brain stimulation.

Beyond the electrodes themselves, the system for signal collection in intracortical stimulation is also crucial for ES.^535,614 Lee et al. unveiled a cutting-edge approach for nerve recording and stimulation, which involves the utilization of wirelessly networked and powered microscale implants (Fig. 23e).⁶⁰³ These small-scale implants are engineered to perform neural monitoring and provide targeted ES autonomously. They communicate bidirectionally with an external hub via a 1 GHz electromagnetic link that works through the skin, enabling personalized control over each device. Research involving 48 neurograins placed on a rat's brain demonstrates this system's advancement in multichannel neural recording and stimulation, offering flexibility for various electrode configurations and scalability for extensive sensor networks. Based on the recording and modulation of brain activity by electrical signals, brain–computer interfaces are designed primarily to improve the life quality for individuals with neurological and sensory disabilities.^{535,615–617} Bidirectional brain-computer interfaces exemplify this integration by translating motor cortex activity into control signals for external devices while also conveying sensory information back to the brain through ES of the somatosensory cortex (Fig. 23f).^595,618 These advanced devices, often implanted directly within or on the surface of the cortex,⁵⁷³ serve as research tools and hold promise for restoring functions by integrating both motor and sensory feedback.^619,620

6.3. Spinal cord stimulation

ES of neural pathways in the spinal cord is facilitated by specialized electrodes designed to deliver precise electrical currents to targeted regions, thereby releasing pain, modulating neurological activity and potentially restoring function impaired by injury or disease.^158,621 SCS involves the placement of electrodes in the epidural space near the dorsal columns of the spinal cord.^622,623 These electrodes generate electric fields that interact with the surrounding tissues, such as the dura and white matter, ultimately activating dorsal column axons and initiating the transmission of action potentials.⁶²⁴ Various stimulation waveforms, including conventional (60–200 Hz), high-frequency (∼10 kHz), and burst stimulation, are employed in SCS to deliver specific patterns of electrical impulses to the tissues.^625–627 Electrodes for SCS, varying in shape and size depending on the application, are meticulously engineered to be biocompatible and to provide stable, controlled stimulation.³⁴ When implanted, they interface directly with neural tissue, forming the critical link between external stimulation devices and the nervous system.⁵⁷⁹ This technology holds promise for enhancing neural plasticity, managing chronic pain,⁵²⁶ and in some cases, reanimating paralyzed muscles by bypassing damaged pathways and directly stimulating the neural circuits below the level of injury.^524,615

Percutaneous linear-type electrodes are thin, flexible leads that are inserted transdermally into the epidural space that encases the spinal cord.^624,628 These electrodes have multiple contact points along their length to deliver ES to specific areas of the spinal cord.⁶²⁹ Percutaneous electrodes are commonly used for temporary or trial stimulations to assess the effectiveness of SCS before considering permanent implantation. Patients undergoing spinal cord therapy will have electrodes placed in the epidural space of the spinal column, which are generally linked to an implanted pulse generator that manages the stimulation settings, as illustrated in Fig. 24a.⁶³⁰ Paddle-type probes are larger than linear-type electrodes, which are implanted surgically through a small incision and placed directly over the spinal cord.⁵²⁷ Paddle electrodes offer a larger contact surface area and are particularly suitable for patients with specific pain patterns or those who require precise targeting of pain areas.⁶³¹ The creation of soft paddle electrodes is designed to tackle the problem of mechanical mismatch between rigid neural implants and the soft neural tissues, which can affect both the durability and functionality of implantable neuroprosthetics in Fig. 24b.⁶³² The e-dura implant is designed to replicate the shape and flexibility of the dura mater and is crafted from a transparent silicone base with stretchable Au wiring, pliable Pt–silicone coated electrodes, and a fluidic channel for drug delivery. Constructed using soft lithography and covalent bonding, the e-dura is durable against numerous mechanical stretches, electrical pulses, and chemical injections. In animal studies, it has shown promise in restoring movement following spinal cord injuries via electrochemical neuromodulation.

Fig. 24 Spinal cord stimulation. (a) Photographs of a spinal implant and a 3D model obtained from high-resolution microcomputed tomography scans taken five weeks after the implant.⁶³⁰ Reproduced from ref. 630 with permission from Springer Nature, copyright 2016. (b) A cross-section image depicts the vertebral column implanted with a soft device in the spinal subdural space, accompanied by an image of the e-dura implant in the same location in rats, and an optical image alongside the SEM image of the Au film and Pt-silicone composite.⁶³² Reproduced from ref. 632 with permission from AAAS, copyright 2015. (c) The images depict a conceptual cross-section of a device implanted in the spinal cord and visualization of the device in both its rolled and unrolled states within the vertebral column.⁶³³ Reproduced from ref. 633 with permission from AAAS, copyright 2021. (d) Diagrams illustrate two distinct neural pathways through which SCS might affect peripheral microcirculation.⁶³⁴ Reproduced from ref. 634 with permission from Elsevier, copyright 1999. (e) Epidural lumbar SCS is shown to create significant rhythmic activity in lower limbs during assisted treadmill stepping and to enable full weight-bearing standing.⁶³⁵ Reproduced from ref. 635 with permission from John Wiley and Sons, copyright 2016. (f) A body weight support system is described that facilitates overground walking, connected to a wireless, implantable pulse generator operating in a closed loop with a paddle lead targeting the dorsal roots serving the lumbosacral segments.⁶³⁶ Reproduced from ref. 636 with permission from Springer Nature, copyright 2022. (g) ES is administered to the spinal cord using two or three percutaneously implanted leads with 8 or 16 contacts each near the lateral lumbosacral area.⁵⁰ Reproduced from ref. 50 with permission from Springer Nature, copyright 2022. (h) A proposed clinical modification entails subcutaneously implanting a device with a PDMS-based telemeter featuring up to ten microchannels, designed to amplify and filter signals to monitor bladder fullness and detect spontaneous emptying, positioned on the dorsal surface of the S1–S2 spinal cord.⁵²⁸ Reproduced from ref. 528 with permission from AAAS, copyright 2013.

The linear type is simpler to implant but tends to be less effective and more susceptible to shifting within the body. In contrast, the paddle-type provides more precise targeting and greater stability, although its implantation necessitates a more invasive surgical procedure.⁶³⁷ The linear electrodes are limited by their spherical electric field and energy inefficiency, while paddle electrodes can be implanted with more electrodes, allowing for more precise and power-efficient stimulation. Recent advances in bioelectronics are revolutionizing SCS devices by incorporating semiconductor techniques to create high-resolution, conformable microelectrode arrays that can include drug delivery systems and are more adaptable than current semi-rigid technologies. The development of a minimally invasive paddle-type SCS (MI-SCS) device that can be percutaneously implanted and then expanded in situ merges the benefits of both types, reducing the invasiveness of neurosurgery while enhancing device performance as shown in Fig. 24c.⁶³³

Developed from the gate control theory introduced by Wall and Melzack in 1965, SCS has advanced as a therapeutic approach for chronic pain conditions, notably failed back surgery syndrome and various neuropathic pains.⁶³⁸ The growing prevalence of neuropathic pain and the drawbacks of prolonged opioid therapy highlight the need for alternative treatments of SCS, which modulates pain perception by activating large fibers in the spinal cord's dorsal columns, triggering action potentials that alter the electrochemical properties of neurons and influence pain signaling both locally and in higher brain centers.⁶²² For ischemia pain, the underlying pain relief may be related to the rebalancing of oxygen need and supply. In this regard, SCS can influence peripheral microcirculation through two different spinal pathways. At moderate intensities, SCS suppresses efferent sympathetic activity, mainly via α1-adrenoreceptors, which causes blood vessels to relax and improves circulation. At higher intensities, SCS can activate an antidromic pathway, potentially triggering the release of a calcitonin gene-related peptide (CGRP) through a different mechanism that may include the activation of inhibitory circuits from the brain projecting to the dorsal horn (Fig. 24d).⁶³⁴ This effect is likely mediated via fast A-delta fibers, resulting in the release of the vasodilatory substance CGRP. This dual pathway mechanism allows SCS to modulate peripheral microcirculation and contribute to pain relief in various intractable pain conditions.⁶³⁴

Besides pain relieving, SCS is also used for achieving functional movements in patients with severe spinal cord injuries.^525,527,630 As epidural SCS can have an impact on the neural circuitry of the distributed propriospinal system by enhancing the circuitry excitation state, it can activate otherwise “silent” translesional volitional motor control. Fig. 24e illustrates the effects of epidural lumbar SCS on lower limb activity when doing stepping on a treadmill and standing with the full weight beared.⁶³⁵ The EMG activity in the lower limbs in assisted stepping with and without SCS shows that SCS significantly enhances EMG and recruits the muscle that can’t provide response to proprioceptive feedback when with only stepping. Additionally, the kinematic representation in the figure demonstrates the transition from sitting to standing, which is assisted by 15 Hz supra-threshold intensity epidural SCS, showing that the subject could start to stand and maintain balance while bearing their full body weight in SCS applied conditions.

By stimulating the spinal cord below the level of injury, SCS aims to enhance sensory input, promote motor function, and potentially improve bladder and bowel control.⁶²¹ After a severe spinal cord injury (SCI), the disruption of signals from the brain results in permanent paralysis because the neurons in the lumbar spinal cord, which are crucial for walking, become nonfunctional. Epidural ES has shown to reactivate these neurons, allowing individuals with paralysis to walk; when combined with neurorehabilitation, it can lead to sustained improvement in walking even without active stimulation.⁵²⁷ Kathe et al. implemented a combination of a surgically implanted neurostimulator, a multi-electrode paddle lead, and a robotic support system to enable a group of individuals with chronic spinal cord injuries to regain functional abilities as shown in Fig. 24f.⁶³⁶

Lateral lumbosacral SCS enables individuals with lower-limb amputations to perceive sensations as if they are coming from their missing foot, which helps restore somatosensory feedback, enhances balance and gait stability, and substantially reduces phantom limb pain. A closed-loop system was developed that modulates stimulation intensity based on real-time signals from a pressure-sensitive insole, resulting in meaningful functional improvements and pain reduction for three individuals with transtibial amputation as shown in Fig. 24g.⁵⁰ These results suggest that lumbosacral SCS is a viable intervention for enhancing somatosensory feedback and reducing pain in lower-limb amputees, warranting further clinical evaluation.

A major issue following SCI is the loss of voluntary bladder control,⁶³⁹ leading to involuntary voiding and coordination problems between bladder contraction and sphincter relaxation.^640,641 To address this, Chew et al. developed a closed-loop neuroprosthetic system in rats (Fig. 24h) that assesses bladder fullness and prevents involuntary voiding without necessitating a dorsal rhizotomy.⁵²⁸ This system used microchannel electrodes implanted in the spine for signal amplification and noise suppression, which recorded sensory information from teased dorsal roots indicating bladder fullness. The neuroprosthetic employed a high-frequency (20 kHz, 7.5 V) block to prevent involuntary voiding and a low-frequency (30 Hz) ventral root stimulator to allow controlled bladder emptying. This approach suggested that combining sensory input from dorsal roots with ventral root stimulation could lead to an effective bladder neuroprosthetic for managing neurogenic bladder without the drawbacks of current surgical interventions.

It's important to note that spinal neural stimulation is typically considered after other conservative treatment options have been exhausted. The specific application and suitability of SCS depend on the individual's condition, pain profile, and response to previous treatments. A comprehensive evaluation by a healthcare professional specializing in pain management or neurology is necessary to determine if spinal neural stimulation is appropriate and likely to be beneficial for a particular patient.

6.4. Vagus nerve and peripheral nerve stimulation

Neuromodulation therapies, including VNS and PNS, utilize electrical impulses to control nerve activity, which can influence immune response, organ functions and muscle control, thus broadening therapeutic options.^533,642 VNS is a therapeutic intervention that has gained significant attention for its potential applications in various medical conditions. The vagus nerve is integral to autonomic functions like regulating heart and respiratory rates, digestion, and immune responses, owing to its mixed composition of motor and sensory fibers and its widespread connections to various body systems.^532,533 First used by JL Corning in the early 1880s to treat epilepsy, VNS has evolved considerably and is currently an FDA-approved therapy for drug-resistant epilepsy and depression in patients.⁶⁴³ VNS plays a crucial role in autonomic regulation, making it a valuable therapeutic option for a variety of conditions, including epilepsy,⁶⁴⁴ seizures,⁶⁴⁵ anti-inflammatory⁶⁴⁶ and modulating sensations through VNS.^647,648 PNS involves the delivering of ES to peripheral nerves or nerve branches outside the CNS for modulation of other organ or muscle functions.^7,11

Vagal circuitry serves as a bidirectional communication pathway between the central and peripheral nervous systems, playing a vital role in regulating inflammation and maintaining homeostasis (Fig. 25a).⁶⁴⁹ The vagus nerve, comprising both afferent and efferent fibers, connects the brainstem to multiple organs, facilitating the transmission of information and allowing the brain to modulate immune responses. Beyond its applications in treating epilepsy and depression, VNS is being explored as an anti-inflammatory therapy due to its capacity to influence the cholinergic anti-inflammatory pathway.^51,650,651 VNS facilitates a feedback loop between the vagus nerve and the brain, modulating cytokine expression to exert anti-inflammatory effects and potentially restoring autonomic balance by reducing pro-inflammatory cytokines such as IL-6, TNFα, and IL-1β, offering potential therapeutic benefits for conditions like cardiovascular disease, arthritis, and Alzheimer's disease⁶⁴⁹

Fig. 25 Vagus nerve stimulation. (a) Demonstration of the vagal circuitry that connects the central and peripheral nervous systems.⁶⁴⁹ Reproduced from ref. 649, originally published by and used with permission from Dove Medical Press Ltd. (b) Anatomical schematic illustrating the cervical and abdominal branches of the vagus nerve.⁶⁵² Reproduced from ref. 652 with permission from Creative Commons Attribution License (CC BY), copyright 2019. (c) Description of the cholinergic anti-inflammatory pathway utilized in sepsis therapy.⁶⁵¹ Reproduced from ref. 651 with permission from Taylor & Francis, copyright 2020. (d) The outer ear and its predominant cutaneous nerves: the auricular branch of the vagus nerve, the greater auricular nerve, and the auriculotemporal nerve.^653,654 (e) Potential pathways through which stimulation of the auricular branch of the vagus nerve may induce cardiovascular effects via stimulation of the nucleus tractus solitarii.⁶⁵³ Reproduced from ref. 653 with permission from Elsevier, copyright 2016.

Studies have shown that VNS aids in rebalancing parasympathetic and sympathetic nervous system activities, effectively diminishing inflammation and helping to halt the progression of conditions like sepsis.⁶⁵¹ Current research is exploring the effects of VNS on multiple inflammatory conditions such as rheumatoid arthritis, Crohn's disease, irritable bowel syndrome, and fibromyalgia, highlighting its broad therapeutic potential for managing inflammation. Inflammatory bowel disease (IBD) represents a chronic gastrointestinal condition that typically manifests in early adulthood and currently has no cure.^655,656 The vagus nerve, which connects the brain to the gut, has been discovered to possess anti-inflammatory effects through two mechanisms: activating the hypothalamic-pituitary-adrenal axis and initiating the cholinergic anti-inflammatory pathway.^657,658 VNS has demonstrated potential in reducing intestinal inflammation and improving gastrointestinal function in IBD, based on animal studies that show improvements in disease activity indices and inflammation markers. Sun et al. highlighted that chronic abdominal VNS could enhance stool quality, lower plasma C-reactive protein levels, and decrease intestinal inflammatory cell populations, effectively reducing chemically-induced inflammation in rats (Fig. 25b).⁶⁵² Cervical VNS modestly reduced inflammatory markers, but key indicators of inflammatory disease, such as the disease activity index, histological evaluations, and cytokine production, did not revert to baseline levels in treated tissues.⁶⁵⁸ Traditional cervical VNS presents clinical challenges and can lead to potential side effects. Payne et al. developed an innovative implantable device for stimulating the subdiaphragmatic vagus nerve, designed to enhance therapeutic efficacy while reducing side effects, demonstrating significant reduction in chemically-induced inflammation in experiments.⁶⁵² Another potential use of VNS is in regulating inflammation and rebalancing the autonomic nervous system's sympathetic and parasympathetic divisions in the case of sepsis (Fig. 25c).⁶⁵¹ Sepsis is marked by an uncontrolled host response to infection, resulting in organ dysfunction. Studies using experimental models have demonstrated that external stimulation of the vagus nerve can regulate inflammation and reestablish the sympatho-vagal balance, potentially enhancing outcomes in sepsis.⁶⁵¹

Transcutaneous VNS (tVNS) offers a promising non-invasive alternative to invasive VNS for various medical conditions.⁶⁴² tVNS is an affordable and straightforward technique, ideally administered at the ear due to its rich sensory innervation, which includes the vagus nerve's auricular branch, along with the auriculotemporal nerve and the greater auricular nerve. (Fig. 25d).^653,654 While optimal sites for tVNS are still being studied, the cymba conchae within the concha is recognized as particularly suitable due to the significant presence of the auricular branch of the vagus nerve. tVNS also impacts heart rate and cardiovascular function through vagal regulation. The stimulation of the auricular branch of the vagus nerve induces cardiovascular effects by activating the nucleus tractus solitarii, which in turn stimulates the dorsal vagal nucleus and the nucleus ambiguus located in the medulla (Fig. 25e).⁶⁵³ This activation enhances cardiovagal outflow while reducing sympathetic outflow to the intermediolateral cell column of the spinal cord. Furthermore, stimulating the NTS could activate the caudal ventrolateral medulla, which then may inhibit the rostroventrolateral medulla, consequently reducing sympathetic nervous system activity.

There are some types of electrodes commonly used for vagus and peripheral nerve stimulation.⁵³¹ Surface electrodes are non-invasive electrodes placed on the skin surface over the targeted peripheral nerve or muscle. They are typically adhesive patches with conductive surfaces that enable electrical current to pass through the skin and stimulate the underlying nerves. Surface electrodes are commonly used in nerve conduction studies, EMG, and transcutaneous electrical nerve stimulation (TENS) therapy.

Both the vagus nerves and peripheral nerves have a similar narrow and elongated anatomy, making them accessible for such interventions.¹⁰ The contemporary implantable VNS device, like the Livanova© stimulator, features a compact, battery-operated stimulator, which is surgically implanted and configured to deliver ES to the vagus nerve, typically encircling the left cervical vagus.⁶⁵⁹ Cuff electrodes, shaped like tubes to encircle peripheral nerves, facilitate a snug fit for efficient nerve stimulation. They consist of an insulated tube with conductive contacts on the inside and are implanted surgically to deliver ES for nerve modulation. A newly developed flexible neural clip implant, as depicted in Fig. 26a, offers a wireless, easily attachable interface for remote nerve modulation, with applications in bioelectronic medicine.⁶⁶⁰ Designed to target nerves such as the vagus and branches of the bladder pelvic and sciatic nerves, the flexible neural clip has shown success in regulating physiological functions in rat models, affecting heart rate, bladder control, and muscle contraction. Its paper clip-like design ensures easy, secure attachment and stable contact with the nerve, improving its surgical usability and effectiveness in therapeutic interventions. Fig. 26b presents the development of flexible split ring electrodes, designed to enhance neural interfaces for advanced prosthetics and bioelectronic medicine by offering reliable recording and stimulation of the sciatic nerve.⁶⁶¹ Their structure prioritizes ease of implantation and minimizes pressure on the nerve, enabling selective muscle activation and superior signal recording. These flexible split ring electrodes, offering improved signal-to-noise ratios compared to traditional cuff electrodes, show significant promise for future applications in neuromodulation.

Fig. 26 Bioelectronics for vagus nerve and peripheral nerve stimulation. (a) Photomicrographs show small vagus and peripheral nerves in rats alongside a schematic detailing how the flexible neural clip modulates nerve functions to control outputs from different organs or tissues. (i) VNS to regulate heartbeat, (ii) stimulation of nerves controlling the bladder to alleviate bladder issues, and (iii) activation of branches of the sciatic nerve to control leg muscle movements.⁶⁶⁰ Reproduced from ref. 660 with permission from John Wiley and Sons, copyright 2017. (b) A schematic illustrates a flexible split ring electrode implanted on the sciatic nerve for precise stimulation.⁶⁶¹ Reproduced from ref. 661 with permission from Elsevier, copyright 2017. (c) Description of an electrode array that is both stretchable and capable of self-rolling.³⁰ Reproduced from ref. 30 with permission from American Chemical Society, copyright 2024. (d) Schematics depicting the growth stages of the sciatic nerve, where electrodes retain its configured shape even when stretched to 50% and 100% strain and the force is subsequently released.⁶⁶² Reproduced from ref. 662 with permission from Springer Nature, copyright 2020. (e) A schematic presenting the conceptual approach to PNS neuromodulation intended to restore motor and physiological functions (left), the interface between the electrode and the nerve (middle), and images of a twining electrode implanted on the vagus nerve (right).⁵³ Reproduced from ref. 53 with permission from AAAS, copyright 2019.

Fig. 26c addresses the challenge of designing self-rolled neural interfaces that can effectively match the complex shapes and movements of nerves.³⁰ This study presents a novel approach utilizing stretchable, flexible electronics crafted from liquid metal–polymer conductors and fabricated through microfluidic printing, capable of showing extreme stretchability (over 600%) and sustained biocompatibility for long-term use. These self-rolled neural interfaces allow for precise, real-time monitoring of neural activity and can tightly conform to dynamic nerve structures, promising significant advancements in neurological diagnostics and therapy.

To address the issues faced by traditional bioelectronic devices, which are unsuitable for young patients with rapidly growing tissues, morphing electronics have been introduced. Liu et al. synthesized a series of viscoelastic electrodes and strain sensors, adapted to the growth of nerve tissue, reducing mechanical stress and the need for repeat surgeries (Fig. 26d).⁶⁶² Their self-healing capabilities enabled chronic ES and monitoring, making them a promising growth-adaptive technology for pediatric electronic medicine. Furthermore, Fig. 26e shows another example of a 3D twining electrode, which employed stretchable mesh wires mounted on a shape memory substrate, utilizing 2D processing techniques.⁵³ This innovative design enabled the electrode to shift from 2D to 3D forms, modulate its stiffness, and return to its original shape at body temperature, thereby gently encircling nerves. Demonstrated in vivo on vagus and sciatic nerves, this electrode showed promise for clinical uses like heart rate modulation and neural signal recording. The selection of an electrode for peripheral neural stimulation is influenced by the specific needs of the medical application, the desired stimulation characteristics, and the overall therapeutic objectives.

PNS and VNS have emerged as revolutionary approaches in the treatment of various diseases and the elicitation of sensations. By delivering targeted electrical impulses to peripheral nerves or the vagus nerve, these therapies can modulate nerve activity, offering relief from conditions such as chronic pain, epilepsy, and depression. VNS, in particular, has been shown to influence neurotransmitter levels and neural circuits associated with mood and seizures. In rehabilitation, peripheral stimulation can restore sensation and muscle control in damaged limbs, bridging the gap between injury and recovery. These techniques can not only alleviate symptoms but also evoke sensory feedback, enhancing the quality of life for patients with sensory deficits. Through precise intervention, PNS and VNS stand at the forefront of a new era in personalized medicine, where the modulation of electrical signals in the body can lead to profound therapeutic benefits.

7. ES based bioelectronics for other applications

ES has become a versatile tool in biofunctional applications, offering a bridge between electronic systems and biological functions. By delivering targeted electrical impulses, this technology can be employed to regulate heart rhythms in pacemakers, facilitate muscle rehabilitation, control acute pain, diminish tremors associated with neurological disorders, and modulate sensory feedback. Its capacity to interface with the body's nervous system allows for precise modulation of biological processes, thus holding the promise of improved quality of life for individuals with various medical conditions. This adaptability underscores the potential of ES to revolutionize treatment paradigms across a spectrum of therapeutic areas.

7.1. Pacemakers

Pacemakers, through the delivery of electrical impulses to the heart muscle, have become a critical treatment for various arrhythmic conditions. They provide targeted ES to correct irregular cardiac rhythms, addressing severe issues such as ventricular fibrillation as well as common conditions like atrial fibrillation and bradycardia. Recent advancements in pacemaker technology include the integration of flexible electronic components, which conform more naturally to the heart's contours, reducing tissue irritation and increasing the device's biocompatibility. This harmony with the body's tissues not only enhances patient comfort but also reduces the risk of device-related complications. These improvements mark a significant step forward in the evolution of cardiac pacing therapy.

Cardiomyocytes naturally produce electrical impulses that initiate in the sinoatrial node and travel through the heart's conduction system, ultimately leading to coordinated muscle contractions that pump blood throughout the body.⁶⁶³ Disruptions in this conduction pathway can cause arrhythmias like tachycardia, atrial fibrillation, and ventricular fibrillation, which may present symptoms ranging from palpitations to sudden cardiac death. In the case of conduction system failure, artificial pacemakers are implanted to restore and maintain a healthy heart rhythm, adapting to physiological needs.^664,665 Advancements in soft electronic technology enhance pacemaker integration with cardiac tissue due to improved biocompatibility and mechanical flexibility, ensuring that the device operates in harmony with the heart's movements.^39,666

Implantable pacemakers require materials that ensure stability and biocompatibility for long-term patient care within the body's challenging biochemical environment. Hwang et al. introduced a novel single-device platform capable of in situ cardiac activity diagnosis and therapeutic ES to prevent arrhythmias, even under wet conditions, without disrupting the recording signal. This device features a pressure-sensitive transistor array for detecting heartbeats, biocompatible electrodes for stimulation, and a hydrogel adhesive for attachment to the epicardium, offering detection and effective pacing for arrhythmia without electrical interference (Fig. 27a).²⁵ Parylene coatings are favored for pacemaker encapsulation due to their biocompatibility and resistance to bodily fluids,^{25,664,665,667} while silicone-based polymers are also utilized for their chemical inertness.^54,668 Electrodes, crucial for delivering ES to the heart, are often made of noble metals like Pt^25,664 and Au,⁵⁴ chosen for their excellent biocompatibility, electrical conductivity, and inertness. Typical hydrogels are also chosen for pacemakers to increase adhesion, prevent tissue inflammation and ensure efficient signal transmission.³¹⁵ These materials are integral to the pacemaker's reliability, enabling it to continuously regulate cardiac rhythm without adverse biological reactions.

Fig. 27 Pacemaker. (a) This bioelectronic device is affixed to the epicardium using an adhesive hydrogel patch that enables recording of cardiac pressure and provides electrical therapy (left), which comprises a pressure-sensitive, active-matrix transistor array and biocompatible pacing electrodes, all secured by encapsulation layers (right).²⁵ Reproduced from ref. 25 with permission from AAAS, copyright 2022. (b) A wireless and biodegradable pacemaker was implanted in an ex vivo Langendorff-perfused mouse model. Upon ES by the pacemaker, synchronous ventricular activation was restored, as shown by the activation map pinpointing the origin at the contact pad. The device's bioresorption post-therapy is also detailed in the images.³⁹ Reproduced from ref. 39 with permission from Springer Nature, copyright 2021. (c) Schematic illustrations and operational principle of an implantable pacemaker based on a PENG.⁶⁶⁷ Reproduced from ref. 667 with permission from American Chemical Society, copyright 2019. (d) Illustration of a symbiotic cardiac pacemaker system based on a TENG.⁵⁴ Reproduced from ref. 54 with permission from Springer Nature, copyright 2019. (e) The pacemaker is wirelessly powered using NFC. The device features the integration of the array onto its body, accompanied by a photographic image showing the array installed on the heart.⁶⁶⁴ Reproduced from ref. 664 with permission from AAAS, copyright 2022.

Biodegradable pacemakers are a groundbreaking advancement in temporary cardiac pacing, removing the necessity for surgical device removal and lowering the risk of associated infections.^669,670 These pacemakers typically employ PLGA as a core material due to its favorable degradation properties in the body, breaking down into metabolically benign lactic and glycolic acids.⁶⁷¹ Mg is also used, particularly for electrodes, because of its biocompatibility and adjustable degradation rates, ensuring that the device dissolves predictably after its therapeutic purpose is served. The electrical properties of Mg facilitate the effective transmission of pacing signals, crucial for managing heart rhythm.⁶⁷² Choi et al. demonstrated a fully implantable, bioabsorbable cardiac pacemaker made of PLGA and Mg, which naturally dissolves in the body post-treatment, highlighting the device's temporary yet therapeutic utility (Fig. 27b).^39,666

The success of implantable pacemakers relies heavily on stable power sources, such as traditional internal batteries like lithium–iodine or lithium–silver vanadium oxide cells, which eventually require surgical replacement once depleted.⁶⁷³ To overcome this limitation, research has turned towards autonomous energy harvesting, including the use of the piezoelectric effect that converts biomechanical energy from cardiac motion into electricity, potentially providing an endless power supply.^667,674,675 Li et al. developed a pacemaker utilizing a piezoelectric composite, Pb(Mg_1/3Nb_2/3)O₃–PbTiO₃, which generates current from heartbeats, eliminating the need for external power (Fig. 27c).⁶⁶⁷ Alternatively, TENGs have shown promise in converting mechanical energy into electrical power.^54,668,676 Ouyang et al. developed an implantable pacemaker powered by a TENG, which not only efficiently harvested and stored energy from cardiac motion but also successfully corrected sinus arrhythmia and prevented deterioration, showcasing its potential for in vivo symbiotic bioelectronic applications (Fig. 27d).⁵⁴ Building on this, Liu et al., from the same research group, introduced a new version of a capsule-shaped intracardiac pacemaker for arrhythmia treatment, which utilizes the coupled effects of triboelectricity and electrostatic induction. This device can be catheter-delivered to the right ventricle of a swine, effectively converting cardiac motion into electricity and maintaining pacing function over a three-week period, thereby demonstrating significant progress in addressing the energy limitations of implantable medical devices.²⁹ RF technology has been used to wirelessly transfer energy, directly powering pacemakers or recharging them.^{39,664–666,677} These innovations, such as Ausra et al.'s miniaturized, subcutaneous pacemaker powered by RF energy harvesting, represent a significant leap towards eliminating battery dependence and enhancing the longevity and convenience of cardiac pacing devices (Fig. 27e).⁶⁶⁴

7.2. Pain block

Pain is a complex bodily response with both sensory and emotional components, triggered by harmful stimuli. It can manifest as acute pain, which is a direct response to injury, or chronic pain, which persists beyond the typical healing period due to various factors.⁶⁷⁸ Opioids, while effective for severe pain relief, carry risks of side effects, dependency, and in some cases, resistance or fatal outcomes, leading to growing interest in alternative pain management methods.⁶⁷⁹ Local electrical nerve stimulation, often referred to as “nerve blockade,” provides a promising drug-free option for acute pain relief by stopping the propagation of action potentials in peripheral nerves. The application of electrical therapy for pain alleviation has ancient roots and was scientifically reinforced in the 1960s by Melzack and Wall, who offered a physiological explanation for its effectiveness.⁶³⁸ This method is widely utilized to relieve pain in various medical conditions, including labor, dysmenorrhea, angina, acute facial pain, dental procedures, and chronic musculoskeletal disorders (Fig. 28a).⁶⁸⁰ ES therapy can be implemented using both non-invasive and invasive electrodes, allowing for flexible application depending on patients' needs and specific medical conditions.^681,682

Fig. 28 Pain block. (a) Commonly targeted locations for ES used in blocking pain.⁶⁸⁰ (b) Hypothesized modes of action for analgesia induced by TENS.⁶⁸⁰ Adapted from ref. 680 with permission from Elsevier, copyright 2009. (c) Three different TENS approaches: conventional TENS, acupuncture-like TENS, and intense TENS.⁶⁸² Redrawn from ref. 682. (d) The schematic diagram of the PENS technique for pain management therapy.⁶⁸³ Reproduced from ref. 683 with permission from Commons Attribution License 4.0 (CC BY), copyright 2021. (e) PENS electrodes. (i) Conventional PENS electrodes are positioned close to the nerve, (ii) increasing the electrical current may activate a larger portion of nerve fibers, and (iii) percutaneous open-coil electrodes can remotely activate a significant proportion of sensory fibers, excluding pain fibers.⁶⁸⁴ Originally published by ref. 684 and used with permission from Dove Medical Press Ltd., copyright 2021. (f) A schematic showing a stimulator with mechanically interwoven connections to tripolar-configured electrodes on a nerve cuff, featuring a bioresorbable design that induces a nerve conduction block and dissolves after two months.³⁸ Reproduced from ref. 38 with permission from AAAS, copyright 2022.

TENS is a non-invasive technique that uses a portable pulse generator and adhesive pads affixed to the skin to manage various types of pain, with its effectiveness influenced by the proximity of the electrodes to the target nerve.^680,682,685 TENS alleviates pain through multiple mechanisms: low-intensity TENS triggers paraesthesia that relieves pain via segmental mechanisms, while high-intensity TENS activates extrasegmental pain inhibitory pathways and blocks peripheral afferent impulses. (Fig. 28b).⁶⁸⁰ In the segmental pain-block mechanism, TENS reduces ongoing nociceptor activity and CNS sensitization when applied to somatic receptive fields, even after spinal cord transection.⁶⁸² The A-delta activity induced by TENS leads to the long-term deactivation of central nociceptors for up to 2 hours. In the extrasegmental mechanism, TENS activates the midbrain periaqueductal grey and the rostral ventromedial medulla, which are parts of the descending pain inhibitory pathways, particularly effective when muscles are stimulated. At the peripheral level, TENS generates antidromic impulses that can collide with and extinguish pain impulses originating from peripheral sites, akin to a “busy-line effect,” typically mediated by large-diameter nerves. Additionally, TENS works through the mediation of neurotransmitters or neurochemicals such as opioids, serotonin (5-HT), acetylcholine, noradrenaline, and GABA, by affecting certain receptors.⁶⁸¹ TENS operates in three distinct modes: conventional, acupuncture-like (AL-TENS), and intense TENS (Fig. 28c).⁶⁸² The International Association for the Study of Pain (IASP) defines conventional TENS as utilizing high-frequency (50–100 Hz) and low-intensity currents with brief pulse widths (50–200 μs) to elicit paraesthesia without discomfort. The primary aim of conventional TENS is to engage large-diameter A-beta afferents, inhibiting nociceptive neurons in the CNS and providing strong but comfortable paraesthesia whenever in pain. Alternatively, AL-TENS uses low-frequency (2–4 Hz), high-intensity pulses with longer pulse widths (100–400 μs) to target A-delta afferents, recommended less frequently and typically used three times daily for 20 minutes, which can be described as strong and comfortable muscle contractions.⁶⁸⁶ Intense TENS, suitable for minor medical procedures, employs very high frequencies (up to 200 pulses per second) with tolerable high intensities for short durations to block pain and facilitate extrasegmental pain relief; however, it would be felt like “maximum tolerable paraesthesia”. Over time, the effectiveness of intense TENS may decrease due to habituation or increased pain, necessitating adjustments in electrode placement, TENS settings, or periodic breaks from the therapy. To summarize, conventional TENS activates large-diameter, non-noxious afferents (e.g., A-beta fibers) to induce segmental analgesia. AL-TENS penetrates deeper, causing muscle twitches and inducing extrasegmental analgesia through the activation of smaller afferents (e.g., A-alpha to A-delta fibers). Intense TENS targets very deep tissues and directly stimulates small, noxious afferents (e.g., A-delta fibers) to block peripherally generated pain. Although different frequency and intensity combinations are used in these three TENS modes, the specific relationship between pulse frequency and outcomes has not been clearly established. Given that individual patients have unique thresholds and preferences, it is recommended to conduct a trial run before TENS treatments to identify appropriate and personalized intensity/frequency parameters that avoid discomfort and best alleviate pain.

When placing TENS electrodes, ensure that they are applied to healthy, sensate skin. Position the electrodes on relevant dermatomes to direct paraesthesia into the painful area, except in cases of hyperaesthesia, mechanical allodynia, hypoaesthesia, and absent or damaged body parts. In such instances, place the electrodes along the main nerves proximal to the pain, paravertebrally at spinal segments, or at contralateral mirror sites. For large or multiple pain areas, use dual-channel TENS devices with four electrodes. Avoid placing electrodes on the anterior neck, over the eyes, through the chest, or internally, except with devices specifically designed for dental, vaginal, or anal use.

Percutaneous electrical nerve stimulation (PENS) is an invasive form of pain management therapy that entails the insertion of fine needles through the skin to target specific nerve pathways directly (Fig. 28d).⁶⁸³ This technique is particularly useful for patients who require deeper nerve stimulation that cannot be effectively reached by transcutaneous methods.^687–689 PENS employs electrical currents to stimulate large-diameter myelinated afferent peripheral nerve fibers, which may reduce the transmission of pain signals from smaller-diameter fibers to the CNS, aligning with the “gate control theory” of pain.⁶⁹⁰ Additionally, PENS is thought to activate descending pain inhibition pathways, enhancing its effectiveness in pain management. ES more effectively activates larger diameter nerve fibers at lower intensities, allowing for the precise targeting of these fibers while minimizing the activation of smaller diameter nociceptive fibers.²⁰⁶ Conventional PENS systems often use small electrodes placed close to the nerve, leading to rapidly decaying electric fields that can result in uneven stimulation; this configuration tends to activate fibers near the electrode, including the less desirable small diameter fibers.⁶⁹¹ In contrast, systems designed for remote selective targeting can activate a higher proportion of large diameter fibers while minimizing stimulation of nociceptive afferents (Fig. 28e).⁶⁸⁴ Such systems utilize leads with large monopolar electrodes placed several centimeters away, generating broad, homogeneous electric fields that can activate large diameter fibers throughout the entire nerve cross-section without activating smaller fibers, thereby reducing unintended discomfort.^684,692 Research indicates that PENS can improve conditioned pain modulation, reduce motor-evoked potentials, and increase intracortical inhibition, indicating that it may be particularly beneficial for patients experiencing central sensitization.^693,694 This modality is employed to treat various types of pain conditions, including chronic musculoskeletal pain, neuropathic pain, and post-surgical pain.⁶⁹⁵

Compared with PENS, invasive cuff-shaped nerve electrodes attach directly to peripheral nerves and deliver stimulation that blocks nerve conduction effectively and rapidly, usually in less than 10 ms, by inducing a sustained depolarization state, thus inhibiting pain signal transmission more efficiently.^696–698 Eggers et al. tested a new carbon-coated electrode that combines direct and kilohertz frequency alternating current to reduce side effects associated with electrical nerve blocks.⁶⁹⁶ Despite their effectiveness, these electrodes require surgical implantation and removal, which carries the risk of additional pain and potential permanent nerve damage. Moreover, implanting non-bioabsorbable, rigid cuffs may cause neuroinflammatory responses and morphological alterations like fibrous encapsulation, complicating the safe extraction of the device.⁶⁹⁹ Fully bioabsorbable peripheral nerve stimulation electrodes were developed to avoid secondary surgical procedures for electrode removal. Lee et al. developed innovative bioabsorbable cuff-shaped electrodes for peripheral nerve stimulation, eliminating the need for secondary surgical removal (Fig. 28f).³⁸ These electrodes, made by slowly dissolving Mo, provide stable pain relief over several weeks, meeting clinical demands for temporary pain management. The Mo electrode, enhanced by a conductive wax coating for added robustness and encapsulated in a bioabsorbable polyanhydride polymer for easy external power connection, offers an effective and direct nerve contact for pain relief. The design ensures that the electrode is gradually absorbed by the body, negating the need for a follow-up surgical procedure.³⁸

ES of peripheral nerves is a non-drug approach to pain management that uses low-voltage currents to activate nerve fibers, blocking pain signals to the brain and reducing pain perception. Despite their differences, all above methods present alternatives to traditional pain management, aiming to reduce reliance on medications such as opioids.⁷⁰⁰ Although effective in pain relief, the details of their underlying mechanisms are not fully understood. This lack of complete understanding hinders the advancement of electrode design for more efficient clinical use.

7.3. Tremor reduction

Tremor is an involuntary, rhythmic oscillation of muscle, resulting in tremulous movement. This phenomenon is most frequently observed in the hands, although it may manifest in other parts of the body as well.^701–703 Tremors can occur sporadically or can be constant, and they can vary in severity. They are often classified by their appearance and cause: resting tremors occur when a muscle is relaxed, such as when hands are lying on the lap, and action tremors occur during the voluntary movement of a muscle. Common in various neurological disorders, tremors are frequently associated with conditions such as Parkinson's disease, multiple sclerosis, and essential tremor, a condition characterized primarily by tremor alone.^559,702 Tremors can also result from certain medications or be a symptom of another underlying medical issue. The tremor is generated by a central brain oscillator, which sends signals to motor neurons directly through the corticospinal tract and indirectly through interneurons. In managing essential tremor, traditional pharmacological and surgical interventions, though primary, often come with considerable side effects, reduced long-term effectiveness, or new clinical contraindications.⁷⁰⁴ ES of nerves is emerging as a promising alternative treatment.⁷⁰⁵ This neuromodulation approach, which applies a mild electrical current to the peripheral nervous system, aims to reduce or stop tremor activity in the limbs. The DBS in target brain regions, such as subthalamic stimulation, can also reduce the tremor activity in Parkinson's disease.

Tremor reduction through peripheral ES employs two main strategies: functional ES (FES), which creates forces in tremor-producing muscles to mechanically reduce tremors, and afferent pathway stimulation, which decreases motoneuron excitability, thereby interrupting the transmission of tremor signals from the CNS to the muscles.⁵⁴⁸ Various strategies for tremor management include employing FES to activate antagonistic muscle forces, stimulating afferent fibers to decrease motor neuron excitability, and activating cutaneous afferent fibers to suppress interneurons, thereby disrupting the propagation of signals that cause tremors (Fig. 29a).⁷⁰⁵

Fig. 29 Tremor reduction. (a) The underlying mechanism by which ES can mitigate tremors involves the disruption of tremorgenic signals. (i) The signals originate from a central brain oscillator and reach motor neuron pools directly through the corticospinal tract or indirectly through interneurons. (ii) applying FES directly to nerves or muscle bellies to activate opposing muscle forces and enhance joint stiffness; (iii) stimulating afferent fibers to reduce the excitability of opposing motor neuron pools, thereby modifying the pathway of tremorgenic signals; (iv) targeting cutaneous afferent fibers to suppress interneurons, which in turn affects the disynaptic transmission of these tremor-inducing signals.⁷⁰⁵ (b) Tremorgenic bursts, indicated by gray lines in the EMG recording window, are detected and future bursts are anticipated during the stimulation window, marked by red and blue lines; accordingly, stimulation is applied out-of-phase to the antagonistic muscles, represented by red and blue rectangles, within this window.⁷⁰⁵ Reproduced from ref. 705 with permission from Creative Commons CC BY license, copyright 2021. (c) Electrodes were positioned on the subject's wrist to target the median and radial nerves, while the counter electrode was placed on the wrist's posterior surface, accompanied by representative spiral images before and after treatment and sham stimulation.⁷⁰⁶ Reproduced from ref. 706 with permission from John Wiley and Sons, copyright 2018. (d) While patients used their dominant hand to draw spirals on a digital tablet, local field potentials were recorded from the bilateral subthalamic nuclei using a ‘wide’ bipolar montage.⁷⁰⁷ Reproduced from ref. 707 with permission from Creative Commons CC BY license, copyright 2024.

FES represents a therapeutic modality of neuromuscular ES designed to facilitate or rehabilitate functional muscle activity. This is achieved through the administration of electrical pulses at an intensity surpassing the motor threshold—the lowest current intensity required to elicit muscle contractions. FES devised position electrodes adjacent to targeted muscles or nerves on the epidermis. Concurrently, a control unit modulated the parameters of the electrical stimuli, including intensity, frequency, and pulse duration, to induce muscle contractions that emulate physiological muscle function.^708–710 Dosen et al. predicted tremor bursts in agonist-antagonist muscle pairs by analyzing EMG signals, then utilized out-of-phase ES to disrupt tremor patterns (Fig. 29b).^701,705 However, this out-of-phase stimulation approach maintains synchronous stimulation with the mechanical tremor frequency, sidestepping EMG artifacts but fails to activate afferent pathways in sync with the actual tremor phase,^701,711 while some studies have employed continuous stimulation without synchronization to specific mechanical or physiological events.⁷⁰²

Afferent pathway ES offers a therapeutic strategy distinct from traditional FES by targeting sensory rather than motor pathways to alleviate tremor. This novel approach applied sub-motor-threshold electrical pulses to tremor-affected dermal areas, activating sensory nerve fibers that convey signals to the CNS, which may modulate motor control regions of the brain and therefore reduce tremor severity.^{702,706,712,713} Lin et al. introduced a non-invasive neuromodulation technique aimed at diminishing hand tremors by applying ES to the wrist median and radial nerves. In their blinded study with twenty-three participants, the active treatment cohort demonstrated a significant tremor attenuation compared to the placebo group, indicating that this preliminary methodology could provide a clinically significant relief for tremor symptoms (Fig. 29c).⁷⁰⁶ Such innovations aim to discern intentional movements from tremorous activity with precision, and to selectively administer stimulation to limbs exhibiting more pronounced tremor symptoms.

In addition to stimulating peripheral nerves, DBS targeting the STN is a recognized and effective treatment for managing tremors. However, its effects on complex everyday movements are not as well understood. Bange et al. conducted a study with Parkinson's patients who performed both template-guided and free spiral drawing tasks while undergoing invasive electrophysiological recordings. The study revealed that beta bursts related ES reduced duration and amplitude of hand tremor, whose changes are particularly evident during free drawing (Fig. 29d).⁷⁰⁷ This research underscores that DBS increases drawing velocity and improves clinical function.

7.4. Sensory modulation

ES, as both a therapeutic modality and an enhancement technology, introduces novel avenues for broadening the spectrum of human sensory experiences. This technique leverages precise electrical signals to strategically activate or modulate designated pathways within the nervous system, thereby simulating the nerve conductions that occur in innate physiological activities.⁷¹⁴ The applications hold promise for the amelioration or reconstitution of various sensory functions, encompassing auditory, visual, olfactory, taste, and tactile modalities.^715,716

7.4.1. Auditory. ES has significantly advanced the field of hearing, particularly with cochlear implants, which are a groundbreaking solution for individuals with severe to profound sensorineural hearing loss (Fig. 30a(i)).^717–719 These devices bypass nonfunctional hair cells by transforming sound signals into electrical impulses that stimulate the cochlea's neural elements. This process enables sound perception in the brain, effectively restoring hearing in affected individuals.⁷²⁰ The choice of electrode materials for cochlear implants is critical due to their extended interaction with the cochlea's sensitive interior and the need for lasting integration within the body.⁷²¹ Materials are selected for their biocompatibility, efficient signal transmission, durability, and minimal interference with the surrounding tissues. Electrode arrays are usually made of durable alloys like TiN, Ti–Ir, titanium–tantalum, IrO_x, and Pt–Ir (Fig. 30a(ii)).^20,722,723 Dalrymple et al. demonstrated that electrodeposited Pt–Ir coated electrodes outperformed the uncoated Pt ones in long-term cochlear implantation and stimulation in rodents, highlighting the advantages of enhanced electrode features, including reduced impedance, higher CSC, and elevated limits for charge injection.⁷²⁴

Fig. 30 Auditory, visual, smell and taste modulation by ES. (a) Auditory modulation. (i) A standard contemporary cochlear implant system transforms sound into electrical impulses, which are subsequently relayed to the auditory nerve.⁷¹⁷ Reproduced from ref. 717 with permission from Springer Nature, copyright 2014. (ii) Cochlear Contour™ electrode arrays.⁷²² Reproduced from ref. 722 with permission from Springer Nature, copyright 2019. (iii) SEM of the assembly with the scaffold structure for drug delivery and flexible electrode array for ES.⁷²⁵ Reproduced from ref. 725 with permission from John Wiley and Sons, copyright 2019. (b) Visual modulation involves a range of techniques for developing a visual prosthesis. (i) Electrodes can be placed in epiretinal, subretinal, or suprachoroidal locations to stimulate retinal ganglion cells, creating visual perceptions known as phosphenes. Additionally, the axons of these cells may be targeted with cuff-style electrodes along the optic nerve, while deeper structures like the lateral geniculate nucleus can be reached using conventional DBS electrodes or more advanced devices featuring a cluster of microelectrodes. Direct stimulation of the visual cortex is possible through the use of either surface or penetrating microelectrodes, enhancing the potential routes for ES of the visual system.⁷²⁶ Reproduced from ref. 726 with permission from Elsevier, copyright 2015. (ii) The illustration highlights different ES pathways in the eye, with a red circle indicating the choroid (except for the eye's anterior portion), a gray semicircle representing the retinal pigment epithelium, and a yellow semicircle depicting the retina itself.⁷²⁷ Reproduced from ref. 727 with permission from Elsevier, copyright 2016. (iii) A color fundus photograph shows a microelectrode array consisting of 60 Pt electrodes implanted in a patient with choroideremia, positioned to efficiently stimulate the retina by resting directly on its surface.⁷²⁸ Reproduced from ref. 728 with permission from Elsevier, copyright 2016. (iv) An enlarged view of one side of the optic self-opening intraneural electrode reveals its active area, complete with flaps shaped 3D, electrodes, and alignment bars, designed to prevent overly deep insertion into the nerve.⁷²⁹ Reproduced from ref. 729 with permission from Springer Nature, copyright 2019. (c) Electrodes are positioned within the nasal concha for smell modulation.⁷³⁰ Reproduced from ref. 730 with permission from Springer Nature, copyright 2018. (d) The schematic and picture for stimulating digital taste involves a detailed tongue interface that connects directly with the participant's tongue.⁷³¹ Reproduced from ref. 731 with permission from Springer Nature, copyright 2018.

Cochlear implantation often faces challenges such as inflammatory responses, fibrosis, and ossification, which can increase electrode impedance and disrupt neural function.^732–735 To reduce mechanical trauma, there is a significant push for softer, more flexible electrode arrays, such as the straight electrode array for round window insertion reported by Lenarz et al., which showed promising insertion dynamics and reduced impact on cochlear structures.⁷³⁶ Soft electrodes and the topical application of steroids are used to preserve hearing function post-implantation, with innovative designs like Jang et al.'s corticosteroid-releasing micro-scaffold showing potential to improve hearing by consistently reducing threshold shifts in guinea pigs (Fig. 30a(iii)).⁷²⁵ After the implantation of the cochlear implant, the electrical impedance at the electrode–tissue interface increases significantly. This elevation in impedance is primarily attributable to the encasement of the electrode array by fibrous tissue.⁷³⁷ This impedance upsurge can be ameliorated even years post-implantation through the strategic intraoperative administration of corticosteroids into the scala tympani, which serves to counteract the fibrotic tissue response within the cochlear environment.⁷³⁸ Luo et al. developed a PCL coating for electrodes that releases dexamethasone to diminish inflammation after surgery, maintaining electrode functionality.⁷³⁹ To combat post-implantation tissue growth, Paasche et al. explored the use of nanostructures on electrode surfaces, resulting in lower impedance and delayed impedance increases, indicating a promising avenue for in vivo applications.⁷⁴⁰

Advancements in cochlear implant technology have included improved insertion techniques for gentler surgery, increased electrode numbers, and refined stimulation protocols.^720,741 However, the field still faces significant challenges, such as optimizing electrode arrays to improve the precision of stimulating cochlear spiral ganglion cells. Current electrode designs can cause suboptimal stimulation due to signal overlap and reduced auditory resolution. Additionally, selecting the appropriate surgical method and electrode type is crucial to avoid damage to cochlear structures, which can impair the implant's effectiveness.

7.4.2. Vision. In visual rehabilitation, ES is a key technique in activating retinal implants that provide signals to neurons under a damaged retina, enabling basic light and shape perception for some visually impaired individuals (Fig. 30b(i)).^726,727 It works by mimicking the role of light in initiating visual perception, allowing signal transmission through the visual pathway despite retinal degeneration.⁷⁴² Additionally, ES targeting the visual pathways and visual cortex is capable of eliciting visual sensations in individuals bereft of sight.⁷⁴³ There are eight main methods of delivering ES, such as transpalpebral, transorbital, and DBS, chosen based on remaining retinal function or, in cases of severe conditions like advanced glaucoma, targeting the brain's visual centers when the retinal pathway is beyond repair (Fig. 30b(ii)).^727,744 Ocular stimulation is less invasive and easier to perform, while cerebral stimulation is reserved for more severe impairments.⁷⁴⁵

Retinal implants using ES are gaining favor in clinical trials due to their ability to tap into the visual pathway's natural processing and their proven effectiveness in restoring functional vision.^728,746,747 The subretinal visual implant has demonstrated significant improvements in the daily activities and mobility of 72% of trial participants over 12 months, with 45% reporting restored visual functions for everyday use.⁷⁴⁷ This implant featured a microphotodiode array on a polyimide substrate, designed to stimulate the retina's bipolar cell layer and utilize the inner retinal neurons’ processing power. Meanwhile, the epiretinal Argus II implant employed a 60-electrode grid to precisely target the retinal ganglion cells, representing another innovative development in visual prostheses (Fig. 30b(iii)).⁷²⁸

Stimulating the optic nerve is an effective approach that bypasses the complex retinal structure directly targeting nerve fibers. This technique preserves the advanced information processing of later visual system stages and is particularly beneficial for severe ocular injuries, such as retinal detachment or globe trauma.⁷⁴⁸ Research has explored various devices for optic nerve stimulation, including spiral sleeve neural electrode arrays^748–751 and penetrating arrays made of Pt–Ir alloys.^752–754 While these devices have been able to induce the perception of phosphenes, their rigidity and mechanical mismatch with neural tissue present challenges for the efficiency of stimulation and long-term implantation. To address these issues, researchers are now using multichannel electrode arrays developed with thin-film microfabrication for optic nerve applications. Gaillet et al. reported an electrode array constructed by layering Ti and Au onto a polyimide substrate, featuring individual stimulation electrodes with an area of merely 0.008 mm⁻² (Fig. 30b(iv)).⁷²⁹ These electrodes exhibited superior mechanical compliance and high charge delivery capacity. Flexible electrodes were laterally implanted into the nerve using a needle and suture method, with the array being anchored within the nerve. Experiments conducted on anesthetized rabbits showed that an intraneural electrode array placed on the intracranial segment of the optic nerve could successfully induce specific activation patterns in the visual cortex upon ES.

ES has witnessed considerable advancements in the realm of vision restoration and the treatment of blindness, with mounting evidence indicating its burgeoning potential in ameliorating a spectrum of visual deficits. The efficacy of ES techniques is contingent upon the meticulous design of electrodes, wherein their geometric configuration and dimensional attributes are critical determinants of the specificity and penetration of stimulation. Optimal electrode deployment is crucial for the efficacious excitation of intended neural substrates, and the resolution of the restored vision is intricately associated with the electrode array's density. Furthermore, the biocompatibility and durability of electrodes are paramount to sustain long-term therapeutic effectiveness and to safeguard patients' health.

7.4.3. Olfaction and taste. Olfaction and taste are critical chemical senses that discern chemical stimuli, aiding in survival by attracting us to nutritious foods and repelling us from dangers like toxins.^647,648 These senses rely on receptors in the nose and tongue to convert a wide range of chemical signals into neural messages our brain understands as distinct smells and tastes. The sensory processing allows us to appreciate the complex aromas and flavors in our environment, highlighting the interconnected nature of taste and smell.⁷⁵⁵

Electrical olfactory stimulation is an incipient technological frontier with the potential to compensate for or amplify olfactory capabilities through the application of electrical currents.^756,757 The olfactory perception process is intricate, encompassing the transduction of odorant molecules into neuronal signals by the olfactory nerve, followed by the transmission of these signals to the brain's olfactory bulb and their subsequent dissemination to various brain regions for intricate processing and identification. Efforts on electrical olfaction aim to mimic this natural process by applying electrical stimuli directly to the olfactory bulb or nerve, which in turn triggers the generation of olfactory signals.⁷⁵⁸ The olfactory receptors, which are sensitive to odor molecules, are situated proximate to the olfactory bulb and the nasal turbinates. The nasal cavity is partitioned into the superior, middle, and inferior turbinates, with the proximity of the superior turbinate to the olfactory receptors rendering it a prime site for electrode placement. Ag electrodes are often strategically positioned here to interface with the olfactory nerve (Fig. 30c).^730,755,759 While empirical research on electrical olfactory stimulation remains nascent, the approach holds theoretical promise for restoring the sense of smell in individuals who have suffered olfactory loss due to pathological conditions, neural impairment, or other etiologies.^{756,760–762}

ES is a burgeoning area of research for mimicking or restoring the sense of taste.^755,763 Ranasinghe et al. were at the forefront with their “Digital Lollipop,” an innovative device that simulates taste by applying ES to the tongue's taste receptors (Fig. 30d).^731,764,765 The device, which features a spherical electrode and a flat base, has been shown to accurately elicit and modulate the four fundamental tastes, sour, salty, bitter, and sweet, at varying intensity and parameters of ES.⁷⁶⁶ Taste sensation can be tuned by various electrical currents.^767,768 Notably, the outcome of the ES was contingent upon the electrode polarity; an anodal electrode near the tongue can induce an electrical or metallic flavor,⁷⁶⁹ whereas a cathodal electrode suppressed the perception of saltiness,⁷⁷⁰ an effect attributed to the intervention of the aqueous electrolyte solution.⁷⁷¹

7.4.4. Haptics. Tactile afferents in the hand transmit details regarding the physical properties of objects and the nature of contact during manipulation, which is critical for everyday activities.⁷⁷² Individuals with diminished tactile sensitivity struggle with these tasks due to the brain's lack of direct mechanical contact information, which cannot be fully compensated by vision or the low sensitivity of proprioceptive afferents to fingertip events.⁶⁰⁵ By delivering controlled electrical impulses to specific neural pathways, ES can provide tactile feedback,^55,56 induce muscle contractions,⁷⁷³ and even restore basic sensations in individuals with sensory deficits, which is important for periprosthetic.²⁷ This technology is pivotal in applications ranging from rehabilitative medical devices and prosthetic limbs to immersive virtual reality experiences, linking the human experience with the digital world in an increasingly interconnected manner.^774,775

Skin electronics and TENS are fascinating aspects of human-machine interface technology that have the capability to induce tactile sensations, effectively tricking the sense of touch. These electrodes are strategically placed on the skin's surface and work by delivering safe, low-level electrical currents that stimulate the cutaneous nerves (Fig. 31a).⁷⁷⁶ The stimulation can mimic diverse tactile sensations, like pressure, vibration, or even texture.⁵⁶ This technology is instrumental in creating sensory feedback in prosthetic devices, allowing users to receive tactile information from their environment. It's also utilized in virtual reality systems to enhance immersive experience, as well as in medical rehabilitation to help patients recover sensory functions. By leveraging the body's own sensory pathways, skin electrodes offer a direct method to enrich human–machine interactions, providing individuals with a more intuitive and responsive experience. Modern prosthetics, inspired by the human body, often lack the tactile feedback crucial for perceiving touch and pain, which are essential for interaction with and protection from our environment.⁶¹⁵ Osborn et al. have used TENS to successfully evoke both nonpainful and painful sensations in a phantom hand, with EEG confirming the activation of the brain's somatosensory regions (Fig. 31b).⁷⁷⁷ The newly developed electronic dermis (e-dermis) mimics biological sensory receptors, allowing amputees to experience a spectrum of tactile stimuli, and has been shown to enable users to distinguish between different tactile sensations and object properties, such as curvature and sharpness.

Fig. 31 Haptic modulation and human machine interfaces. (a) A schematic graph showing the electrotactile process where an electrode is attached on the fingertip.⁷⁷⁶ Reproduced from ref. 775 with permission from Springer Nature, copyright 2022. (b) Sites on the median and ulnar nerves of an amputee's residual limb are targeted, activating specific regions in the phantom hand through TENS.⁷⁷⁷ Reproduced from ref. 776 with permission from AAAS, copyright 2018. (c) By the third week post-operation, cuff electrodes demonstrated high selectivity in evoking sensations, with each contact point on the nerve cuff inducing either a distinct sensation or a sensation in a specific location; each point is identified by a letter indicating the nerve and a number denoting the stimulation channel on the cuff around that nerve.⁷⁷⁸ Reproduced from ref. 777 with permission from AAAS, copyright 2014. (d) The prosthetic limb was connected to an abutment that transferred the load directly to the bone through an osseointegrated fixture.⁷⁷⁹ Reproduced from ref. 778 with permission from AAAS, copyright 2014. (e) Bidirectional control of a hand prosthesis was achieved by adjusting the current based on sensor readouts from the prosthetic hand, with S15 and S75 representing 15% and 75% of the sensor range, respectively. A photograph illustrating the surgical placement of a TIME electrode in the participant's median nerve, alongside an image showing the ulnar nerve with two electrodes implanted.⁷⁸⁰ Reproduced from ref. 779 with permission from AAAS, copyright 2014.

Implanted peripheral nerve interfaces could enable individuals with limb amputations to experience consistent, natural touch sensations in their phantom hands. Tan et al. designed a kind of cuff electrode placed around the nerves to provide ES, which induced tactile perceptions in different areas of the phantom hand. These sensations include tapping, pressure, moving touch, and vibration as shown in Fig. 31c.⁷⁷⁸ These sensations were reported to feel natural and free from tingling or paresthesia. Intensity patterns of the stimulation could adjust the perceptual area size, while changes in frequency affected the sensation strength, improving the subjects’ prosthesis control and object manipulation. The study involved two male subjects who received the implants following industrial accidents, providing 20 stimulation channels for the first subject and 16 for the second. The subjects could reliably report the location and type of sensation in response to the ES of each channel, demonstrating the potential for sustained sensory restoration following limb amputation via non-invasive peripheral nerve cuff electrodes.

The development of implantable devices has long faced the issue of establishing a reliable and long-lasting method for transcutaneous communication, particularly for prosthetic limbs that require intuitive neural control that existing interfaces do not provide. To overcome this, Max et al. have created a percutaneous osseointegrated (bone-anchored) interface that facilitates ongoing, two-way communication with the human body, allowing artificial limbs to be controlled by implanted electrodes in an amputee's nerves and muscles during everyday activities (Fig. 31d).⁷⁷⁹ This interface has been shown to provide a more accurate and consistent control than traditional surface electrodes, working effectively in various limb positions and environmental settings with reduced effort from the user. It enables stable, long-term myoelectric pattern recognition and provides sensory feedback via neurostimulation, improving the prosthetic limb's natural sensation. This technology marks substantial progress in mimicking natural limb functionality, offering stable attachment and continuous, dependable communication between the human body and the prosthetic device.

An ideal prosthetic replacement should provide the user with natural-like sensations when grasping or manipulating objects, which requires the simultaneous and real-time decoding of user intent and delivery of sensory feedback.^618,781 While peripheral nervous system neural interfaces have shown potential, there has been no conclusive evidence of their effectiveness in real-time control of dexterous prosthetic hands. Raspopovic et al. introduced a closed-loop system, where the control signal, derived from the chest's superficial electromyogram, complicates real-time bidirectional control due to the possibility of sensory gating when muscles must be contracted and touch sensations felt simultaneously (Fig. 31e).^780,782 This sensory feedback allowed the test subject to modulate the prosthesis's grasping force accurately without relying on visual or auditory cues and to discern the stiffness and shape of objects, substantially enhancing the prosthesis's life-like functionality and user experience.

Advancements in ES technology are continually expanding the horizons of human sensory exploration, offering the promise of transforming our interaction with the world and enriching our quality of life. Advancements in research and technology are paving the way for a future filled with innovative sensory experiences that are currently beyond our most imaginative expectations.

8. Conclusion and outlook

In this review, we have explored the intricate mechanisms and diverse applications of ES and its associated bioelectronics, highlighting its profound impact on cellular behavior and potential for therapeutic intervention and biofunctional modulation. We have discussed how ES can manipulate the TMP to trigger a wide range of cellular responses, including migration, proliferation, apoptosis, differentiation, and attachment. This manipulation is crucial for deepening our understanding of cellular and molecular dynamics, as well as for leveraging these responses in medical and therapeutic contexts. We have highlighted the versatility of ES in modifying various pathophysiological processes across multiple biological scales, from molecules to organs. ES plays a particularly significant role in fields such as tissue engineering, neurology, cardiology, and sensory system management. It is essential for tissue regeneration, pain management, restoration of motor functions, and rhythm regulation. These typical bioelectronics are summarized in Table 2. However, many mechanisms behind the effects of ES on specific diseases remain unexplained, and some unexpected results have been observed in both research and clinical applications. Additionally, patient variability significantly affects the outcomes of ES treatments, underscoring the need for carefully tailored ES protocols. While ES offers immense potential for therapeutic applications, further research is required to fully understand its mechanisms and optimize its use. By continuously investigating and refining ES protocols, we can further enhance ES efficacy and reliability, ultimately improving patient outcomes in various medical fields.

Table 2 ES based bioelectronics for different biomedical applications

Types	Biomedical application	Device type	ES parameters	ES channels	Encapsulation	Size	Power supply	Other properties	Ref.
Abbreviations: GOx, glucose oxidase; MN, microneedles; PPy, polypyrrole; PVA, polyvinyl alcohol; HA, hyaluronic acid; PAM, polyacrylamide; PU, polyurethane; PA, polyamide; EBFCs, enzymatic biofuel cells; PEG, polyethylene glycol; PET, polyethylene terephthalate; PDMS, polydimethylsiloxane; PEDOT, poly(3,4-ethylenedioxythiophene); PSS, polystyrene sulfonate; PCL, polycaprolactone; PLGA, poly(lactic-co-glycolic acid); SEBS, styrene-ethylene-butylene-styrene; PTFE, polytetrafluoroethylene; MAA, methacrylated alginate; AAm, acrylamide; NIPAM, N-isopropylacrylamide; ITO, indium tin oxide; Au, gold; Ag, silver; NWs, nanowires; Pt, platinum; Ir, iridium; IrOx, iridium oxide; Ti, titanium; Cu, copper; Mg, magnesium; Mo, molybdenum; Al, aluminum; Cr, chromium; Zn, Zinc; Fe, iron; Mn, manganese; W, tungsten; Si, silicon; MoS2, molybdenum disulfide; CNT, carbon nanotube; TENG, triboelectric nanogenerator; PENG, piezoelectric nanogenerator; PLA, polylactic acid; PLLA, poly(L-lactic acid); PLA-AP, poly(l-lactic acid)-block-aniline pentamer-block-poly(l-lactic acid); PI, polyimide; PGMA, poly-glycidyl methacrylate; PTMC, poly(trimethylene carbonate); b-DCPU, bioresorbable dynamic covalent polyurethane; IU, isophorone bisurea; MPU, methylenebis(phenyl urea); Alg-CA, conductive tissue adhesive catechol-conjugated alginate; SMP, shape memory polymer; SF, silk fibroin; STN, subthalamic nucleus; DBS, deep brain stimulation; VNS, vagus nerves stimulation; PNI, peripheral nerve injury; CP, common peroneal; FPCB, flexible printed circuit board; EF, electric field; NFC, near-field communication; VP, viscoplastic; MRI, magnetic resonance imaging; PD, Parkinson's disease; EGaIn, eutectic gallium–indium; LiCl, lithium chloride; PAAm, polyacrylamide, BLE, bluetooth low energy.
Skin healing	Diabetic wound healing	MN	3.53 μA	GOx, horseradish peroxidase	PPy/dopamine/PVA/HA	—	EBFCs	Self-powered	376
	Wound healing	Ionic TENG patch	10 Vpp, 4 Hz	Ionic gel	LiCl/PAM based organic gel	—	TENG	Stretchable, self-powered	392
	Wound healing	Film electronics	15 V, 250 Hz, 100 μs duration	EGaIn	PEG/PDMS	∼150 μm thick	Wired to external equipment	Wet adhesive	394
	Wound healing	Flexible patch	6 V, 10 Hz, 1 ms width	Ag NWs/MAA	Medical patch	∼5 μm thick	—	Printable electrodes	393
	Chronic wound healing	Smart bandage	2 V, 13.56 Hz	PEDOT: PSS/P(AAm/NIPAM) hydrogel	Gel/FPCB	FPCB: ∼100 μm thick	NFC	Adhesive and closed-loop	375
	Infected chronic wound healing	Film device	1 V, 50 Hz	Au	SEBS	∼300 μm thick;	Battery	Closed-loop, combination treatment	379
						∼20.0 mm × 30.0 mm
Muscle regeneration	Biceps femoris muscle regeneration	Film device + MN	Pulse electric field: 1 V, 10 Hz, 50 ms duration	Mg	PLGA	∼130 μm thick; ∼15.0 mm × 30.0 mm	NFC	Biodegradable, drug delivery	49
	Intestinal tissue healing	Film device	Pulse EF: 30 V, 1 ms duration, 0.3 s interval, 100 pulses/D.C. EF: 0–1 V for 3 days	Mo, Mg	PCL, chitosan hydrogel	< 200 μm thick; ∼10.0 mm × 15.0 mm	Mg–Mo galvanic cells	Biodegradable	37
	Inflammation inhibition in tendon injury	Floating-gate memory interdigital circuit	Bionic spikes, average amplitudes of ∼0.175 mA (0.04–0.6 mA)	Cr/Au electrodes, MoS2 semiconductive channels	PI	∼30 μm thick	Wired to external equipment	Biomimetic programmable	254
Bone regeneration	Bone fraction healing	Film device	∼4 Vpp, 1 Hz	Mg	PLGA	∼35 × 15 × 0.45 mm^3	TENG	Biodegradable, self-powered	377
	Osteoblasts’ proliferation and differentiation	Package	60 mV, 1 nA	Au/PTFE/ITO	PDMS/PET	∼20.0 mm × 20.0 mm	TENG	Self-powered	436
	Rabbit radial defect regeneration	Composite scaffold	500 mV, 100 Hz	PLA-AP/PLGA/HA	—	—	Wired to external equipment	Drug delivery	438
	Rat femoral defect regeneration	Hydrogel	500 mV, 100 Hz	Calcium phosphate/PEDOT:PSS/magnesium titanate/MAA	Gel	—	Wired to external equipment	Locally promote calcium influx	437
Nerve regeneration	Axonal regeneration and myelin regeneration	Conductive nanofibers composite catheter	Square anodic pulses: 0.1 ms pulse width, 3 V, 20 Hz, 1 h duration	PPy/SF	SF nanofibers	10 mm length;	Wired to external equipment	Biocompatible	102
						1 mm/1.1 mm, inner/outer diameter
	Myelin regeneration and restoring motor function	Piezoelectric polymer support	—	ZnO	PCL	0.48 mm thick	ZnO/PCL nanogenerator	Self-powered	479
	Long-term acceleration of PNI recovery	Porous nerve conduits and patch	Rhythmic waveform:1.6 ± 0.1 cycle s^−1, 1.1–2.7 V, synchronizing with respiratory motion	PEDOT	Chitosan, PLLA, PDMS, PI	—	Pt wire, TENG/PENG	Self-powered	449
	Regeneration of damaged nerves	Cylindrical conduit wrapped in film	231.4 mW h^−1 cm^−3	Zn, CNT/Pt	PLGA	12 mm in length; 0.5 mm thick	Zn–O2 battery	Self-powered	450
	Nerve regeneration and functional recovery	PLGA cuff and receiver antenna	Monophasic step pulses: 200 μs per pulse, 20 Hz, 1 h	Mo, Mg coil, electrodes, Mg/SiO2/Mg capacitor	PLGA	0.2 mm thick	Wireless RF power supply	Bioresorbable	486
	Promotes nerve regeneration and monitors nerve repair	Conductive nerve conduits	3 V, 20 Hz, 1 h every other day	Mg	PLA	17 mm in length	Mo wire, PENG	Biodegradable, self-powered	452
	Accelerates nerve regeneration and restores function in vivo	Self-electrified conduit device	0.068–0.984 V	Mg, FeMn	Porous PCL and PLLA-PTMC films	∼700 μm thick	Mg–FeMn galvanic cell	Biodegradable, self-electrified	488
	Promotes neuromuscular recovery	PLGA cuff and receiver antenna	Monophasic, square waveform, 200 μs pulse width, 20 Hz, minimum amplitude over threshold, 2–4 V	Mo	PLGA, b-DCPU	∼200 μm thick	Wireless RF power supply	Bioresorbable	378
Drug delivery	Wound healing	Package	∼1.1 V	Cu	PDMS	∼20.0 mm × 40.0 mm	Stretchable battery	Self-adaption	494
	Transdermal Iontophoresis	Patch	0.75 V	Enzyme-modified carbon fabrics	Medical tape	∼5 mm thick, 30.0 mm × 10.0 mm	EBFCs	Self-powered	495
	Transdermal Iontophoresis	Patch	4 V, 12 μA	PET/PTFE/Al	Kapton	10 cm × 10 cm	TENG	Self-powered	783
	Transdermal electroosmotic flow	Porous MN array patch	0.7 V, 0.2 mA	Carbon fabric	PGMA	PMN: 28° angle of ϕ0.13 mm cone, 0.25 mm height; cylindrical supporting post: 300 μm height, 450 μm diameter	EBFCs	Self-powered	497
	Diabetes treatment	Mesoporous MN	0.1 V	Steel/Au	PDMS	4.3 cm^2	Battery	Closed-loop	499
	Wound healing	Patch	15 V	Carbon	PI, PDMS	50.0 mm × 20.0 mm	NFC	Closed-loop	500
	Skin Electroporation	Facial mask	100 V, a pulse width of 200 μs	Carbon	Silicone	—	Wired to external controller	Stretchability	507
	SARS-CoV-2 vaccination	MN patch	30 V, 55 ms	Stainless steel	Plastics	MNE: 650 μm in length, 200 μm × 50 μm in cross-section	Piezocrystal	—	508
	DNA and siRNA delivery	MN chip	35 V, 20 ms	Au	Silicon/parylene	MNE: 190 μm in height, 340 μm in spacing	—	Low voltage, flexibility	509
Brain stimulation	STN-DBS for Parkinson's disease	Lead	100 μs rectangular waveform, −1 to −5 mA (0.5 mA step size), followed by a 5-ms low amplitude anodic stimulation	Pt–Ir alloy	PU outer jacket	1.27 mm in diameter, four electrodes 1.5 mm in length/40 oval shaped electrode contacts	Wired to external equipment	—	551
	Intracortical microstimulation	Flexible thread	0–50 μA	Ti/Pt/IrOx microcontacts	Fused silica	1 μm thick, 36–100 μm in width, 1800 μm in length	Wired to external equipment	Focal and spatially selective neural activation	552
	Chronic brain mapping and modulation	Injectable mesh	5–50 μA, 1 ms pulse duration, pulse trains with interval of 1 s	Pt/Cr	SU-8	∼800 nm thick and∼20 μm wide mesh elements; total mesh width, W = 2 mm	Wired to external equipment	Evoke single-neuron responses to ES	553
	STN-DBS in PD rats	Fiber	Square constant current pulses: biphasic, symmetric, 130 Hz, 50–200 μA, 60 μs width per phase	Graphene	Glue	∼75 μm in diameter	Wired to external equipment	MRI compatibility	554
	Epileptic seizures	Epidural electronics	1 V, 100 Hz, 30 s duration	Graphene	SU-8	∼5.5 μm thick	Wired to external equipment	—	599
	Mouse visual cortex stimulation	Surface arrays integrated with depth electrodes	Constant-current biphasic, 100 μs phase square, 32–60 μA, 2 ms intervals	PEDOT:PSS, Au	Parylene	Circular electrodes: 15 μm in diameter, 40 μm in pitch	Wired to external equipment	—	600
	Sheep motor cortex stimulation	Endovascular stent	Monophasic anodal and cathodal 1 Hz single pulses and pulse trains: 20–50 Hz, 200–300 μs, <10 mA	Pt	Nitinol stent, and PI	Electrodes of 500 μm or 750 μm in diameter, 50 μm thick	Wired to external equipment	Self-expanding	602
Spinal cord stimulation	Epidural electrical stimulation for motoneuron activation	Injectable linear-type implants	0.1–0.3 mA, 20–80 Hz	Au	PI	Lead ∼35.5 mm × 0.8 mm; connector: ∼7 mm × 6 mm	Wired to external equipment	—	630
	Spinal cord injury recovery	Electronic dura mater	40Hz, 0.2ms pulse duration, 50–200 μA	Pt-silicone composite	PDMS	120 μm thick	Wired to external equipment	Drug delivery	632
	Motor and sensory signal elicitation, bridging complete spinal cord injuries	Flexible thin-film electronics	10–200 μA, 50–100 Hz, 10–20 pulse trains	Ti/Au/PEDOT:PSS	Parylene	Device: ∼4 μm thick, 8.90, 9.90, and 10.90 mm in length; Electrode: 0.1 mm × 0.1 mm	Wired to i360 device	—	637
	MI-SCS	Paddle-type device	—	Au	Parylene, silicone	Fluidic layer: 30–60 μm thick; electronic layer: 4 μm thick	Wired to external equipment	Shape-changing	633
VNS/PNS	VNS reducing heart rates; sciatic nerve stimulation	Self-rolled neural interface	VNS: 1 mA, 20 Hz, 500 μs pulse width for rats; 10 V for rabbits; Sciatic nerve: rectangle waves of 50 μA, 100 μA, and 150 μA, 3 Hz, 150 μs pulse width	EGaIn	TPU/PDMS	200–300 μm thick; self-rolled electrodes: 160 μm in inner diameter	Wired to external equipment	—	30
	VNS reducing heart rates	Twining electrodes	0.4 mA, 100 μs wave width, 10 Hz	Au/Ti	PI/SMP	∼100 μm thick; 1 mm wide	Wired to external equipment	—	53
	Vagus nerves: heart rate modulation; bladder nerve: bladder dysfunction; sciatic nerve branch: muscle activation	Neural clip	VNS: 30 Hz, 0.4 ms pulse width, 3.3 s duration, 0.2 mA; Pelvic nerve: 25–200 μA; CP nerve: 500 μs pulse width; tibial nerve: 170 μs pulse width	IrO2/Au	PI	Clip-cavity: 700 μm × 500 μm/470 μm × 200 μm, clip strip: 650 μm × 900 μm/420 μm × 650 μm; clip-string: 700 μm wide/470 μm	Wired/Wirelessly powered by a coil	—	660
	Sciatic nerve stimulation	Split ring	∼200 μA	Au/Pt	PI	Open ringed structure of 1.3 mm inner diameter, tips of 500 μm inner diameter, and the outer diameter of 2.5 mm	Wired to external equipment	—	661
	Sciatic nerve stimulation	Growth-adaptive neural interfaces	Constant voltage pulses with intensity from 0 to 1 V	PEDOT:PSS/glycerol	PDMS-IU-MPU	PEDOT:PSS/glycerol: 2 μm thick; VP substrate: 120 μm thick	Wired to the external stimulator	Viscoplastic and self-healing properties	662
Pacemaker	In situ cardiac activity diagnosis and therapeutic ES	Film device	Pulse amplitude, 0.6 V mm^−1; 5 Hz; pulse width, 1 ms	Pt, Au, Pt black	PDMS, Silastic MDX4-4210, parylene	∼38 μm thick	Wired to the external stimulator	Heart motion monitoring	25
	Cardiac monitoring and electrical stimulation	Film device	1 ms pulse width, peak-to-peak voltage of 2.5 V	EGaIn composite	Alg-CA hydrogel, PDMS	< 200 μm thick	Wired to the external stimulator	Conformable tissue adhesion	315
	Cardiac pacing	Film device with receiver antenna	Transmitting voltage, 1 Vpp; frequency, 10 Hz	W/Mg	PLGA	∼250 μm thick	Wireless RF power supply	Bioresorbable	39
	Transient closed-loop detection and cardiac pacing	Transient closed-loop system	∼100 bpm	W/Mg	b-DCPU, PLGA	∼250 μm thick	Wireless RF power supply	Bioresorbable	666
	Cardiac pacing	Pacemaker and PENG	∼2.5 V, pulse width ∼2 ms, 80 bpm	—	PDMS, parylene	—	PENG	Self-powered	667
	Cardiac pacing	Pacemaker and TENG	∼4 V, pulse width ∼0.9 ms	—	PDMS	—	TENG	Self-powered	54
Pain block	Mixed chronic pain	Needle-electrode monopolar probes	2–100 Hz, 0.2–0.5 V	—	—	Probe: 5 to 20 cm in length	Wired to external device	Ultrasonic guide in insertion process	683
	Electronic pain block	Cuff	10 Vpp, 25 kHz	Mo/Mg	PLGA/PA	Mo: 300 μm wide, 15 μm thick; Mg: 1 mm wide, 50 μm thick; PLGA: ∼40 μm thick; PA: ∼200 μm thick; Complete system: 8 mm wide, 400 μm thick	Wired to external device	Biodegradable	38
Tremor reduction	Essential tremor relief	Wristband	300 μs biphasic pulses, 50 μs interpulse period, 150 Hz	Hydrogel	—	—	—	Noninvasive	706,712
	Inhibition of Parkinsonian tremor	Non-woven surface electrode	200 μs pulse width, 250 Hz pulse frequency	Commercial electrodes	—	—	—	Noninvasive	702
Sensory modulation	Provides auditory clues	Electrode array	—	Pt-Ir	Silicone	0.31–0.37 mm in diameter	Wired to external equipment	—	724
	Provides auditory clues	Micro-scaffold cochlear electrode array	65 μs pulse width, 19.8 Hz, 2–10 VPP	Pt	PI	4.28 mm in length, 270 μm in height, 610 μm in width	Wired to external equipment	Loading drugs	725
	Visual prosthesis therapy	Integrated circuits and microelectrode devices	—	Pt	PI	Electrodes: 3 mm × 2 mm	Wireless RF power supply	Implantable	728
	Visual cortex activation	Ring electrode array	10–2000 μA, 50–400 μs pulse durations	Au	PI	12 μm thick, 33 mm long, 3 mm wide	Wired to external equipment	Mechanical stability	729
	Induces olfaction	Device with long electrodes	—	Ag	—	—	Wired to external equipment	—	759
	Induces taste	Device with lollipop electrodes	50–1200 Hz, 20–200 μA	Ag	—	—	Wired to external equipment	—	764
	Hand tactile feedback	Skin-integrated electronics	Pulse waveform, 25–500 Hz, 0–13.5 mA	LiCl-PAAm hydrogel	PDMS	Driver unit: 19.2 g, 5 cm × 5 cm × 2.1 mm;	Lithium-ion battery	Moisture permeability, BLE wireless communication	776
						Hand patch: 220 μm thick for most of the area, ∼1 mm thick for the hydrogel patches, 11.04 g
	Multimodal haptic feedback	Skin-integrated electronics	Pulse waveform, 0–250 Hz, 40–100V	Au	Thermally conductive silicone, PI	2 × 4 patch: ∼2.8 × 4.0 cm^2;	Lithium-ion battery	BLE wireless communication	56
						4 × 4 patch: ∼4.5 × 7.5 cm^2;
						Palm patch: ∼114 cm^2
	Long-term natural touch perception generation	Peripheral nerve cuffs	Monopolar, biphasic, charge-balanced, cathodic-first pulses, 0–255 μs pulse width, 0–1000 Hz, < 5.6 mA, 50 V	Pt	Silicone sheath	A common anode: 5.08 cm × 10.16 cm on the dorsal surface	Wired to external equipment	Implanted, stable response for 16 and 24 months	778
	Long-term sensory feedback and motor control of artificial limbs	Bone-anchored interfaces	Charge-balanced biphasic pulse	Pt	Silicone	Self-sizing spiral cuff electrodes	Wired to a neurostimulator	Bidirectional communication with the human body	779

---

The future of ES is set to expand with advancements in bioelectronics, materials science, and nanotechnology. Emerging technologies are expected to enhance the precision, efficiency, and adaptability of ES devices. Opportunities lie in the development of more sophisticated and miniaturized ES systems that can seamlessly integrate with the human body, providing real-time feedback and personalized treatment options. Innovative materials and fabrication techniques can lead to more durable and flexible devices, improving patient comfort and compliance.

Looking ahead, future advancements in bioelectronics for ES are expected to benefit greatly from breakthroughs in materials science. The development of biocompatible, flexible, and highly conductive materials will enhance the integration of electronic devices with biological tissues, promising higher efficiency and precision in delivering ES, reducing energy requirements, and improving safety profiles. Moreover, the advent of nanoengineering and microfabrication technologies is poised to transform the design and functionality of bioelectronic devices, enabling the creation of more complex and miniaturized systems capable of interacting at the cellular or even sub-cellular level.

As our understanding of the bioelectrical underpinnings of diseases advances, ES can be customized to individual physiological and pathological conditions, ushering in a new era of precision medicine. Personalized ES therapies could be developed by integrating patient-specific genetic, proteomic, and metabolic data, optimizing therapeutic outcomes and minimizing side effects. This personalized approach will require robust algorithms and models to predict individual responses to ES, necessitating extensive interdisciplinary research involving bioinformatics, systems biology, and computational modeling.

The integration of ES devices with wireless technologies and the Internet of Medical Things could significantly enhance the monitoring and real-time adjustment of therapies. These systems could enable continuous remote health monitoring and dynamic adjustment of ES parameters based on real-time data, improving patient compliance and treatment efficacy. Wireless technologies could also promote the development of closed-loop systems, where sensors and actuators work synergistically to maintain optimal physiological states.

As ES technologies advance and become more popular in routine clinical practices, ethical and regulatory frameworks will need to evolve accordingly. The potential for these technologies to modify neurological functions or other critical biological processes necessitates careful consideration of ethical implications, including issues of consent, privacy, and the potential for misuse. Regulatory bodies will be challenged to establish rigorous standards and guidelines to ensure the safety and efficacy of new ES devices and therapies, balancing innovation with patient safety.

ES stands as a dynamic and versatile tool in bioelectronics, poised to make significant contributions to healthcare and medicine. By continuing to explore the complex interactions between electrical fields and biological tissues, together with advancing the technologies that facilitate these interactions, we can look forward to developing novel therapeutic strategies that are more effective, personalized, and integrated with the body's natural bioelectrical systems. The journey of ES from fundamental research to widespread clinical application offers a promising path toward addressing some of the most challenging medical issues at this time, ultimately enhancing patients' outcomes and quality of life.

Data availability

This review does not include any new research data generated by the authors. All figures presented in this review are used with proper authorization from the original sources. The original sources of all figures can be found in the corresponding captions.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the Research Grants Council of the Hong Kong Special Administrative Region (Grant No. RFS2324-1S03, 11213721, 11215722, and 11211523), the Innovation and Technology Fund of Innovation and Technology Commission (Grant No. ITS/119/22), National Natural Science Foundation of China (Grant No. 62122002), City University of Hong Kong (Grant No. 9667221, 9678274, and 9610444), in part of InnoHK Project on Project 2.2–AI-based 3D ultrasound imaging algorithm at Hong Kong Centre for Cerebro-Cardiovascular Health Engineering (COCHE), and Shenzhen Science and Technology Innovation Commission (Grant no. SGDX20220530111401011).

References

1. J. Liu, et al. , Nat. Nanotechnol., 2015, 10, 629.
2. M. L. Kringelbach, N. Jenkinson, S. L. F. Owen and T. Z. Aziz, Nat. Rev. Neurosci., 2007, 8, 623.
3. S. R. Patel and C. M. Lieber, Nat. Biotechnol., 2019, 37, 1007.
4. J. P. Reilly and A. M. Diamant, Electrostimulation: theory, applications, and computational model, Artech House, 2011.
5. R. Balint, N. J. Cassidy and S. H. Cartmell, Tissue Eng., Part B, 2013, 19, 48.
6. S. M. Won, E. Song, J. T. Reeder and J. A. Rogers, Cell, 2020, 181, 115.
7. T. Gordon and A. W. English, Eur. J. Neurosci., 2016, 43, 336.
8. E. Kim, et al. , Interdiscip. Med., 2023, 1, e20230003.
9. E. Song, J. Li, S. M. Won, W. Bai and J. A. Rogers, Nat. Mater., 2020, 19, 590.
10. J. Yi, et al. , Nature, 2023, 624, 295.
11. C. Russell, A. D. Roche and S. Chakrabarty, Int. J. Intell. Rob. Appl., 2019, 3, 11.
12. C. Chen, X. Bai, Y. Ding and I.-S. Lee, Biomater. Res., 2019, 23, 25.
13. A. J. Hayes and J. Melrose, Adv. Therapeutics, 2020, 3, 2000151.
14. M. N. Rasband, Nat. Rev. Neurosci., 2010, 11, 552.
15. A. S. Rowlands and J. J. Cooper-White, Biomaterials, 2008, 29, 4510.
16. Z. Liu, X. Wan, Z. L. Wang and L. Li, Adv. Mater., 2021, 33, 2007429.
17. K. S. Cole, Membranes, ions and impulses: a chapter of classical biophysics, Univ. of California Press, 1972, vol. 1.
18. F. Rattay, Neuroscience, 1999, 89, 335.
19. D. R. Merrill, M. Bikson and J. G. R. Jefferys, J. Neurosci. Methods, 2005, 141, 171.
20. S. F. Cogan, Annu. Rev. Biomed. Eng., 2008, 10, 275.
21. H. Lee Seung, et al. , Sci. Adv., 2021, 7, eabj4624.
22. J. Gehl, Acta Physiol. Scand., 2003, 177, 437.
23. M. Björninen, et al. , Ann. Biomed. Eng., 2017, 45, 1015.
24. P. Fattahi, G. Yang, G. Kim and M. R. Abidian, Adv. Mater., 2014, 26, 1846.
25. J. C. Hwang, et al. , Sci. Adv., 2022, 8, eabq0897.
26. Q. Zhang, G. Zhao, Z. Li, F. Guo and Y. Huang, Biosens. Bioelectron., 2024, 263, 116597.
27. Q. Zeng and Z. Huang, Adv. Funct. Mater., 2023, 33, 2301223.
28. Q. Yang, et al. , Nat. Mater., 2021, 20, 1559.
29. Z. Liu, et al. , Nat. Commun., 2024, 15, 507.
30. R. Dong, et al. , ACS Nano, 2024, 18, 1702.
31. H. Wu, et al. , Colloids Surf., A, 2023, 676, 132197.
32. K. W. Cho, et al. , Chem. Rev., 2022, 122, 5068.
33. M. J. Nelson, P. Pouget, E. A. Nilsen, C. D. Patten and J. D. Schall, J. Neurosci. Methods, 2008, 169, 141.
34. N. H. Shahemi, et al. , ACS Biomater. Sci. Eng., 2023, 9, 4045.
35. S. Choi, S. I. Han, D. Kim, T. Hyeon and D.-H. Kim, Chem. Soc. Rev., 2019, 48, 1566.
36. S. Jin, et al. , Nature, 2023, 623, 58.
37. H. Wu, et al. , Nat. Electron., 2024, 7, 299–312.
38. G. Lee, et al. , Sci. Adv., 2022, 8, eabp9169.
39. Y. S. Choi, et al. , Nat. Biotechnol., 2021, 39, 1228.
40. X. Zhao, et al. , Nat. Mater., 2024, 23, 703–710.
41. P. Sousa-Victor, L. García-Prat and P. Muñoz-Cánoves, Nat. Rev. Mol. Cell Biol., 2022, 23, 204.
42. S. Sun, I. Titushkin and M. Cho, Bioelectrochemistry, 2006, 69, 133.
43. J. J. B. Jack, D. Noble and R. W. Tsien, Electric current flow in excitable cells, Clarendon press, Oxford, 1975.
44. B. Ferrigno, et al. , Bioact. Mater., 2020, 5, 468.
45. G. Thrivikraman, S. K. Boda and B. Basu, Biomaterials, 2018, 150, 60.
46. X. Huang, et al. , Nano Lett., 2022, 22, 3447.
47. P. Wu, et al. , Adv. Mater., 2024, 36, 2310483.
48. H. Wu, et al. , Giant, 2024, 19, 100300.
49. Y. Huang, et al. , Nano Lett., 2022, 22, 5944.
50. A. C. Nanivadekar, et al. , Nat. Biomed. Eng., 2023 DOI:10.1038/s41551-023-01153-8.
51. L. V. Borovikova, et al. , Nature, 2000, 405, 458.
52. A. L. Hodgkin and A. F. Huxley, Nature, 1939, 144, 710.
53. Y. Zhang, et al. , Sci. Adv., 2019, 5, eaaw1066.
54. H. Ouyang, et al. , Nat. Commun., 2019, 10, 1821.
55. Y. Huang, et al. , Mater. Today Phys., 2022, 22, 100602.
56. Y. Huang, et al. , Nat. Electron., 2023, 6, 1020.
57. O. Bouevitch, A. Lewis, I. Pinevsky, J. P. Wuskell and L. Loew, Biophys. J., 1993, 65, 672.
58. M. Levin, BioEssays, 2012, 34, 205.
59. H. Wu, et al. , Adv. Funct. Mater., 2024, 34, 2309270.
60. Y. Chai, et al. , Biomaterials, 2022, 280, 121310.
61. L. Li, et al. , Mater. Today Adv., 2022, 16, 100301.
62. X. Zhang, et al. , Mater. Today Adv., 2023, 17, 100343.
63. B. Feng, et al. , Bioact. Mater., 2023, 22, 127.
64. S. Coyle, et al. , Acta Biomater., 2022, 142, 149.
65. E.-t Wang and M. Zhao, Chin. J. Traumatol., 2010, 13, 55.
66. J. Zhang, et al. , Adv. Healthcare Mater., 2021, 10, 2000604.
67. J. S. Khaw, R. Xue, N. J. Cassidy and S. H. Cartmell, Acta Biomater., 2022, 139, 204.
68. N. Hlavac, et al. , Ann. Biomed. Eng., 2021, 49, 3401.
69. M. H. Lee, et al. , Cells, 2021, 10, 2846.
70. Z. Zhao, et al. , Stem Cell Res., 2012, 8, 38.
71. Y. Dai, J. Mu and F. Zhou, Rev. Cardiovasc. Med., 2021, 22, 1167.
72. J. Liu, et al. , Biosens. Bioelectron., 2022, 209, 114252.
73. U. H. Ko, et al. , Tissue Eng., Part A, 2018, 24, 752.
74. F. Sahm, et al. , Int. J. Mol. Sci., 2020, 21, 6944.
75. L. Wang, et al. , Chem. Eng. J., 2021, 424, 130563.
76. B. Cortese, I. E. Palamà, S. D'Amone and G. Gigli, Integr. Biol., 2014, 6, 817.
77. L. Yao and Y. Li, Stem Cell Rev. Rep., 2016, 12, 365.
78. J. I. Hoare, A. M. Rajnicek, C. D. McCaig, R. N. Barker and H. M. Wilson, J. Leucocyte Biol., 2016, 99, 1141.
79. A. Kumar, K. Nune and R. Misra, Biomater. Sci., 2016, 4, 136.
80. A. Guo, et al. , J. Invest. Dermatol., 2010, 130, 2320.
81. L. Forciniti, J. Ybarra III, M. H. Zaman and C. E. Schmidt, Acta Biomater., 2014, 10, 2423.
82. A. El-Badawy, et al. , Sci. Rep., 2016, 6, 37801.
83. Z. Zhao, et al. , Eur. Cell Mater., 2011, 22, 58.
84. J.-F. Feng, et al. , Stem Cells, 2012, 30, 349.
85. S. N. Iwasa, R. Babona-Pilipos and C. M. Morshead, Stem Cells Int., 2017, 1, 4276927.
86. H. Zhao, A. Steiger, M. Nohner and H. Ye, PLoS One, 2015, 10, e0129625.
87. B. W. Liu, S. C. Zhang, X. L. Wang, J. Y. Yu and B. Ding, J. Colloid Interface Sci., 2015, 457, 203.
88. M. R. Love, S. Palee, S. C. Chattipakorn and N. Chattipakorn, J. Cell. Physiol., 2018, 233, 1860.
89. S. Mobini, L. Leppik, V. T. Parameswaran and J. H. Barker, PeerJ, 2017, 5, e2821.
90. X. Wang, et al. , Front. Med., 2016, 10, 286.
91. S. H. Llewellyn, et al. , Adv. Biol., 2021, 5, 2000271.
92. M. Griffin, S. A. Iqbal, A. Sebastian, J. Colthurst and A. Bayat, PLoS One, 2011, 6, e23404.
93. D. R. Hipfner and S. M. Cohen, Nat. Rev. Mol. Cell Biol., 2004, 5, 805.
94. M. L. Hernández-Bule, M. A. Trillo and A. Ubeda, PLoS One, 2014, 9, e84636.
95. X. Chen, et al. , PLoS One, 2014, 9, e86421.
96. E. A. Oshin, et al. , Sci. Rep., 2024, 14, 885.
97. F. Ma, et al. , Biosci. Rep., 2016, 36(6), e00420.
98. Q. Wan, et al. , Am. J. Phys. Med. Rehabil., 2016, 95(1), 28–38.
99. B.-S. Guo, K.-K. Cheung, S. S. Yeung, B.-T. Zhang and E. W. Yeung, PLoS One, 2012, 7, e30348.
100. M. Wang, et al. , Exp. Biol. Med., 2013, 238, 951.
101. Y.-M. Zhang, et al. , Clin. Exp. Pharmacol. Physiol., 2007, 34, 742.
102. Y. Zhao, et al. , Biomaterials, 2020, 255, 120164.
103. W. Kim and G. Kim, Bioact. Mater., 2024, 35, 382.
104. Y.-C. Chan, et al. , J. Cardiovasc. Trans. Res., 2013, 6, 989.
105. D. Hernández, et al. , Stem Cells Int., 2016, 1718041.
106. M. Guillot-Ferriols, S. Lanceros-Méndez, J. G. Ribelles and G. G. Ferrer, Biomater. Adv., 2022, 138, 212918.
107. M. Hronik-Tupaj, W. L. Rice, M. Cronin-Golomb, D. L. Kaplan and I. Georgakoudi, BioMed. Eng. OnLine, 2011, 10, 9.
108. L. J. Kobelt, A. E. Wilkinson, A. M. McCormick, R. K. Willits and N. D. Leipzig, Ann. Biomed. Eng., 2014, 42, 2164.
109. H.-F. Chang, Y.-S. Lee, T. K. Tang and J.-Y. Cheng, PLoS One, 2016, 11, e0158133.
110. F. Han, et al. , Adv. Healthcare Mater., 2021, 10, 2100027.
111. H. Cheng, Y. Huang, H. Yue and Y. Fan, Stem Cells Int., 2021, 1.
112. R. Zhu, et al. , Exp. Neurol., 2019, 319, 112963.
113. L. J. Kobelt, A. E. Wilkinson, A. M. McCormick, R. K. Willits and N. D. Leipzig, Ann. Biomed. Eng., 2014, 42, 2164.
114. Z.-y Dong, et al. , Neurosci. Lett., 2017, 651, 109.
115. K.-A. Chang, et al. , PLoS One, 2011, 6, e18738.
116. B. S. Eftekhari, M. Eskandari, P. A. Janmey, A. Samadikuchaksaraei and M. Gholipourmalekabadi, Adv. Funct. Mater., 2020, 30, 1907792.
117. R. Karimi-Soflou, I. Shabani and A. Karkhaneh, Int. J. Biol. Macromol., 2023, 237, 124063.
118. S. Staehlke, et al. , Cells, 2022, 11, 2650.
119. C. Heo, et al. , Biomaterials, 2011, 32, 19.
120. A. Babaie, et al. , Eur. Polym. J., 2020, 140, 110051.
121. L. Jaatinen, et al. , Biointerphases, 2016, 11, 011004.
122. Q. Qiu, M. Sayer, M. Kawaja, X. Shen and J. Davies, J. Biomed. Mater. Res., 1998, 42, 117.
123. X. Liu, Z. Yue, M. J. Higgins and G. G. Wallace, Biomaterials, 2011, 32, 7309.
124. X. Zhang, et al. , Mater. Today, 2023, 68, 177.
125. R. M. Dorrian, C. F. Berryman, A. Lauto and A. V. Leonard, Front. Cell. Neurosci., 2023, 17, 1095259.
126. L. Leppik, K. M. C. Oliveira, M. B. Bhavsar and J. H. Barker, Eur. J. Trauma Emergency Surgery, 2020, 46, 231.
127. K. Kawamura and Y. Kano, Neurosci. Lett., 2019, 698, 81.
128. N. Inoue, et al. , J. Am. Coll. Cardiol., 2004, 44, 914.
129. S. Zhao, A. S. Mehta and M. Zhao, Cell. Mol. Life Sci., 2020, 77, 2681.
130. J. Zhang, M. Li, E.-T. Kang and K. G. Neoh, Acta Biomater., 2016, 32, 46.
131. J. Huang, Z. Ye, X. Hu, L. Lu and Z. Luo, Glia, 2010, 58, 622.
132. Y. Li, et al. , Cell. Signalling, 2019, 59, 141.
133. Y. Li, et al. , Mol. Med. Rep., 2019, 19, 4727.
134. B.-j Lin, et al. , Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 8568.
135. I. Titushkin and M. Cho, Biophys. J., 2009, 96, 717.
136. C. D. Rae, V. H.-C. Lee, R. J. Ordidge, A. Alonzo and C. Loo, Int. J. Neuropsychopharmacol., 2013, 16, 1695.
137. A. Y. Gerasimenko, et al. , Polymers, 2022, 14, 1866.
138. B. Molitoris, A. Geerdes and J. McIntosh, J. Clin. Invest., 1991, 88, 462.
139. J. R. Jacobson, et al. , Am. J. Physiol.: Lung Cell. Mol. Physiol., 2006, 291, L289.
140. S. Mobini, L. Leppik and J. H. Barker, Biotechniques, 2016, 60, 95.
141. N. Zhang, et al. , Biosens. Bioelectron., 2018, 112, 149.
142. D. Janigro, et al. , BMC Cancer, 2006, 6, 1.
143. Y. Tai, et al. , Adv. Healthcare Mater., 2021, 10, 2100806.
144. J. C. Silva, et al. , Sci. Rep., 2024, 14, 5458.
145. R. Maidhof, et al. , J. Tissue Eng. Regener. Med., 2012, 6, e12.
146. V. Hosseini, S. Gantenbein, I. Avalos Vizcarra, I. Schoen and V. Vogel, Adv. Healthcare Mater., 2016, 5, 2045.
147. A. Sesena-Rubfiaro, et al. , ACS Biomater. Sci. Eng., 2023, 9, 1644.
148. P. Li, et al. , Adv. Wound Care, 2023, 12, 498.
149. R. Hess, et al. , Biomaterials, 2012, 33, 8975.
150. Y.-C. Fu, et al. , PLoS One, 2014, 9, e91581.
151. B. Chu, et al. , Nano Energy, 2022, 100, 107471.
152. R. Nazari-Vanani, et al. , Biomater. Adv., 2023, 149, 213364.
153. J. Chen, et al. , Nano Energy, 2023, 106, 108076.
154. C. Zhi, et al. , Adv. Mater., 2024, 2401264.
155. Y. S. Choi, et al. , Nat. Biotechnol., 2021, 39, 1228.
156. Q. Yang, et al. , Nat. Mater., 2021, 20, 1559.
157. S. Wang, et al. , Sci. Adv., 2023, 9, eadj0540.
158. A. Burton, et al. , Nat. Commun., 2023, 14, 7887.
159. X. Xiao, et al. , Small Methods, 2023, 7, 2201350.
160. M. Silverå Ejneby, et al. , Nat. Biomed. Eng., 2022, 6, 741.
161. P. Li, et al. , Nature, 2024, 1.
162. Y. Zhang, et al. , Adv. Healthcare Mater., 2021, 10, 2100695.
163. J. C. Chen, et al. , Nat. Biomed. Eng., 2022, 6, 706.
164. F. Qi, et al. , Composites, Part B, 2022, 237, 109864.
165. E. Kim, et al. , Interdiscip. Med., 2023, 1, e20230003.
166. S. D. Dutta, et al. , Biomaterials, 2023, 294, 121999.
167. S. Shaner, et al. , Lab Chip, 2023, 23, 1531.
168. Q. Zhang, et al. , Microphysiol. Syst., 2018, 2, 4.
169. L. Yildirimer, et al. , Biofabrication, 2019, 11, 032003.
170. H. Zhou, et al. , ACS Nano, 2024, 18, 2077.
171. X. Yuan, D. E. Arkonac, P.-H. G. Chao and G. Vunjak-Novakovic, Sci. Rep., 2014, 4, 3674.
172. L. Sordini, et al. , Front. Bioeng. Biotechnol., 2021, 9, 591838.
173. G. Thrivikraman, et al. , ACS Appl. Mater. Interfaces, 2015, 7, 23015.
174. R. Guo, et al. , Acta Biomater., 2022, 139, 105.
175. S. Wang, et al. , Brain Res., 2023, 1798, 148163.
176. C. M. Tringides, et al. , Adv. Healthcare Mater., 2023, 12, 2202221.
177. S. Lu, et al. , Small Methods, 2023, 7, 2200883.
178. S. Shaner, et al. , Lab Chip, 2023, 23, 4967.
179. H. Wu, et al. , Biofabrication, 2023, 15, 042001.
180. J. W. Kim, et al. , Lab Chip, 2022, 22, 2122.
181. Q. Li, et al. , Chin. Chem. Lett., 2022, 33, 1308.
182. Z. Gilbert, et al. , Clin. Neurophysiol., 2023, 152, 93.
183. D. Yang, Y.-I. Shin and K.-S. Hong, Front. Neurosci., 2021, 15, 629323.
184. C. Boehler, S. Carli, L. Fadiga, T. Stieglitz and M. Asplund, Nat. Protoc., 2020, 15, 3557.
185. D. W. Kumsa, et al. , J. Neural Eng., 2016, 13, 052001.
186. A. M. O’Mahony, D. S. Silvester, L. Aldous, C. Hardacre and R. G. Compton, J. Chem. Eng. Data, 2008, 53, 2884.
187. W. Franks, I. Schenker, P. Schmutz and A. Hierlemann, IEEE Trans. Biomed. Eng., 2005, 52, 1295.
188. V. K. Murthy, T. M. Grove, G. A. Harvey and L. J. Haywood in The Second Annual Symposium on Computer Application in Medical Care, 1978. Proceedings. 610.
189. O. Kwon, et al. , Healthcare Inform. Res., 2018, 24, 198.
190. S. A. Go, K. Coleman-Wood and K. R. Kaufman, J. Electromyography Kinesiology, 2014, 24, 31.
191. S. Vanhatalo, J. Voipio and K. Kaila, Clin. Neurophysiol., 2005, 116, 1.
192. S. S. You, et al. , Nat. Electron., 2024, 7, 497.
193. H. J. Landau, Proc. IEEE, 1967, 55, 1701.
194. P. Bhusal, et al. , Drug Delivery Transl. Res., 2018, 8, 708.
195. S. B. Baumann, D. R. Wozny, S. K. Kelly and F. M. Meno, IEEE Trans. Biomed. Eng., 1997, 44, 220.
196. S. Negi, R. Bhandari, L. Rieth and F. Solzbacher, Biomed. Mater., 2010, 5, 015007.
197. S. Negi, R. Bhandari and F. Solzbacher in 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 5142.
198. J. Bobacka, A. Lewenstam and A. Ivaska, J. Electroanal. Chem., 2000, 489, 17.
199. I. C. Stefan, et al. , J. Electrochem. Soc., 2001, 148, E73.
200. M. Łukaszewski, M. Soszko and A. Czerwiński, Int. J. Electrochem. Sci., 2016, 11, 4442.
201. N. Elgrishi, et al. , J. Chem. Educ., 2018, 95, 197.
202. M. Ganji, A. Tanaka, V. Gilja, E. Halgren and S. A. Dayeh, Adv. Funct. Mater., 2017, 27, 1703019.
203. Z. Aqrawe, et al. , Polymers, 2020, 12, 1654.
204. A. L. Hodgkin and A. F. Huxley, J. Physiol., 1952, 117, 500.
205. R. V. Shannon, IEEE Trans. Biomed. Eng., 1992, 39, 424.
206. D. T. Brocker and W. M. Grill in Handbook of Clinical Neurology, ed. A. M. Lozano and M. Hallett, Elsevier, 2013, vol. 116, p. 3.
207. D. F. Williams, Annu. Rev. Mater. Sci., 1976, 6, 237.
208. E. M. Thaning, M. L. M. Asplund, T. A. Nyberg, O. W. Inganäs and H. von Holst, J. Biomed. Mater. Res., Part B, 2010, 93B, 407.
209. V. Woods, et al. , J. Neural Eng., 2018, 15, 066024.
210. C. Boehler, F. Oberueber, S. Schlabach, T. Stieglitz and M. Asplund, ACS Appl. Mater. Interfaces, 2017, 9, 189.
211. A. Vanhoestenberghe and N. Donaldson, J. Neural Eng., 2013, 10, 031002.
212. S. Arrhenius, Z. Phys. Chem., 1889, 4, 96.
213. P. Čvančara, et al., bioRxiv, 2019, preprint DOI:10.1101/653964.
214. P. H. Chan, M. Yurko and R. A. Fishman, J. Neurochem., 1982, 38, 525.
215. X. S. Zheng, C. Tan, E. Castagnola and X. T. Cui, Adv. Healthcare Mater., 2021, 10, 2100119.
216. H. Yuk, J. Wu and X. Zhao, Nat. Rev. Mater., 2022, 7, 935.
217. A. Domínguez-Bajo, et al. , Biomaterials, 2021, 279, 121186.
218. Y.-W. Hu, et al. , Adv. Sci., 2024, 11, 2307746.
219. C. Boehler, D. M. Vieira, U. Egert and M. Asplund, ACS Appl. Mater. Interfaces, 2020, 12, 14855.
220. L. Wang, et al. , ACS Nano, 2023, 17, 22277.
221. L. Kaya, et al. , Adv. Sci., 2023, 10, 2301854.
222. B. Yuan, et al. , Acta Biomater., 2022, 139, 82.
223. R. Sait, et al. , Electrochim. Acta, 2024, 144527.
224. C.-H. Fan, et al. , ACS Nano, 2023, 17, 9140.
225. H. Yu, et al. , Nano Energy, 2024, 121, 109225.
226. J. Leal, N. Jedrusik, S. Shaner, C. Boehler and M. Asplund, Biomaterials, 2021, 275, 120949.
227. I. Perkucin, et al. , Adv. Healthcare Mater., 2022, 11, 2201164.
228. P. Wu, et al. , Front. Bioeng. Biotechnol., 2020, 8, 709.
229. C. Dong, F. Qiao, W. Hou, L. Yang and Y. Lv, Appl. Mater. Today, 2020, 21, 100870.
230. J. Li, X. Liu, J. M. Crook and G. G. Wallace, Mater. Sci. Eng., C, 2020, 107, 110312.
231. Y. Zhao, et al. , Int. J. Nanomedicine, 2024, 19, 2341.
232. X.-P. Li, et al. , Colloids Surf., B, 2021, 205, 111844.
233. R. Liu, et al. , Appl. Surf. Sci., 2021, 560, 149965.
234. N. Zheng, et al. , ACS Nano, 2022, 16, 2292.
235. Y. Huang, W. Jing, Y. Li, Q. Cai and X. Yang, Colloids Surf., B, 2021, 204, 111785.
236. N. Driscoll, et al. , Sci. Transl. Med., 2021, 13, eabf8629.
237. L. Mao, et al. , Adv. Healthcare Mater., 2020, 9, 2000872.
238. T. Wang, et al. , Sci. Adv., 2024, 10, eadi6799.
239. W. Liu, et al. , Adv. Healthcare Mater., 2023, 12, 2301817.
240. D. Jung, et al. , Science, 2021, 373, 1022.
241. R. A. Frederick, I. Y. Meliane, A. Joshi-Imre, P. R. Troyk and S. F. Cogan, J. Neural Eng., 2020, 17, 056001.
242. S. Liu, et al. , Appl. Surf. Sci., 2021, 545, 148827.
243. I. Shaheen, et al. , J. Energy Storage, 2023, 72, 108719.
244. R. Atmaramani, et al. , Acta Biomater., 2020, 101, 565.
245. S. F. Cogan, P. R. Troyk, J. Ehrlich, T. D. Plante and D. E. Detlefsen, IEEE Trans. Biomed. Eng., 2006, 53, 327.
246. S. Venkatraman, et al. , IEEE Trans. Neural Syst. Rehabilitation Eng., 2011, 19, 307.
247. R. D. Meyer, S. F. Cogan, T. H. Nguyen and R. D. Rauh, IEEE Trans. Neural Syst. Rehabil. Eng., 2001, 9, 2.
248. T. Sun, et al. , Acta Biomater., 2023, 159, 394.
249. J. Haas, et al. , Sens. Mater., 2020, 32, 2903.
250. S. F. Cogan, P. R. Troyk, J. Ehrlich and T. D. Plante, IEEE Trans. Biomed. Eng., 2005, 52, 1612.
251. R. Green, et al. , J. Neural Eng., 2013, 10, 016009.
252. A. Ahnood, A. Chambers, A. Gelmi, K.-T. Yong and O. Kavehei, Chem. Soc. Rev., 2023, 52, 1491.
253. M. Pandeeswar, H. Khare, S. Ramakumar and T. Govindaraju, Chem. Commun., 2015, 51, 8315.
254. R. Bao, et al. , Nat. Commun., 2024, 15, 1327.
255. Y. Wu, et al. , ACS Appl. Mater. Interfaces, 2019, 11, 4833.
256. M. Canillas, B. Moreno, M. Carballo-Vila, J. Jurado and E. Chinarro, Mater. Sci. Eng., C, 2019, 96, 295.
257. K. P. Olczak and K. J. Otto in Electrochemical Society Meeting Abstracts prime, 2020, 1576 (The Electrochemical Society, Inc.).
258. Y. Jiang and B. Tian, Nat. Rev. Mater., 2018, 3, 473.
259. C. Qin, et al. , Appl. Mater. Today, 2020, 21, 100804.
260. S. Lee, B. Ozlu, T. Eom, D. C. Martin and B. S. Shim, Biosens. Bioelectron., 2020, 170, 112620.
261. M. Jia and M. Rolandi, Adv. Healthcare Mater., 2020, 9, 1901372.
262. A. Maziz, E. Özgür, C. Bergaud and L. Uzun, Sens. Actuators, Rep., 2021, 3, 100035.
263. N. Rossetti, P. Kateb and F. Cicoira, J. Mater. Chem. C, 2021, 9, 7243.
264. N. Almufleh, A. Al-Othman, Z. Alani, M. H. Al-Sayah and H. Al-Nashash, Electronic Materials, 2021, 2, 413.
265. S. Manchineella, et al. , Adv. Healthcare Mater., 2016, 5, 1222.
266. M. Nune, S. Manchineella, T. Govindaraju and K. S. Narayan, Mater. Sci. Eng., C, 2019, 94, 17.
267. B. Maity, S. Alam, S. Samanta, R. G. Prakash and T. Govindaraju, Macromol. Biosci., 2022, 22, 2200097.
268. M. A. Bhat, R. A. Rather and A. H. Shalla, Synth. Met., 2021, 273, 116709.
269. X. Fan, et al. , Adv. Sci., 2019, 6, 1900813.
270. G. Dijk, H. J. Ruigrok and R. P. O'Connor, Adv. Mater. Interfaces, 2021, 8, 2100214.
271. C. Boehler and M. Asplund in 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 2202.
272. J. Rivnay, et al. , Nat. Commun., 2016, 7, 11287.
273. S. Santhanam, et al. , Adv. Mater. Technol., 2023, 8, 2201724.
274. K. Wang, et al. , Adv. Mater., 2019, 31, 1805867.
275. M. Naguib, V. N. Mochalin, M. W. Barsoum and Y. Gogotsi, Adv. Mater., 2014, 26, 992.
276. G. Basara, et al. , Acta Biomater., 2022, 139, 179.
277. Y. Fu, J. Zhang, H. Lin and A. Mo, Mater. Sci. Eng., C, 2021, 118, 111367.
278. J. S. Kim, et al. , ACS Nano, 2023, 17, 14706.
279. W. Wu, et al. , Chem. Eng. J., 2022, 427, 131988.
280. A. Serafin, M. Culebras, J. M. Oliveira, J. Koffler and M. N. Collins, Adv. Compos. Hybrid Mater., 2023, 6, 109.
281. L. Agrawal, S. K. Vimal, P. Barzaghi, T. Shiga and M. Terenzio, Macromol. Biosci., 2022, 22, 2200315.
282. A. F. Rodrigues, et al. , Nanoscale, 2023, 15, 687.
283. H. Nekounam, et al. , Surf. Interfaces, 2023, 40, 102926.
284. H. P. Bei, et al. , Molecules, 2019, 24, 658.
285. N. Anzar, R. Hasan, M. Tyagi, N. Yadav and J. Narang, Sens. Int., 2020, 1, 100003.
286. V. Sivasubramaniyam, S. Ramasamy, M. Venkatraman, G. Gatto and A. Kumar, Energies, 2023, 16, 3665.
287. J. Cao, Q. Wang and H. Dai, Phys. Rev. Lett., 2003, 90, 157601.
288. Y.-C. Yang, et al. , ACS Appl. Nano Mater., 2021, 4, 7917.
289. J. S. Meena, S. B. Choi and J.-W. Kim, Electron. Mater. Lett., 2022, 18, 256.
290. I. C. Lee, Y.-C. E. Li, J. L. Thomas, M.-H. Lee and H.-Y. Lin, Mater. Horiz., 2024, 11, 876.
291. F. Qi, et al. , Mater. Chem. Front., 2023, 7, 1671.
292. H. Lin, Y. Chen and J. Shi, Adv. Sci., 2018, 5, 1800518.
293. K. Huang, Z. Li, J. Lin, G. Han and P. Huang, Chem. Soc. Rev., 2018, 47, 5109.
294. N. Driscoll, et al. , Sci. Transl. Med., 2021, 13, eabf8629.
295. L. Peng, N. Abbasi, Y. Xiao and Z. Xie, Adv. Mater. Interfaces, 2020, 7, 2001538.
296. Y. Zhang, C. Ma, J. Xie, H. Ågren and H. Zhang, Adv. Mater., 2021, 33, 2100113.
297. L. Zeng, et al. , Chemistry, 2022, 8, 632.
298. P. Wu, et al. , Adv. Mater., 2024, 2310483.
299. C. Xu, et al. , Adv. Funct. Mater., 2020, 30, 2000177.
300. W. Liu, et al. , Adv. Healthcare Mater., 2023, 12, 2301817.
301. J. H. Koo, et al. , Nat. Electron., 2023, 6, 137.
302. D.-H. Kim, et al. , Science, 2011, 333, 838.
303. C. Labouesse, et al. , Nat. Commun., 2021, 12, 6132.
304. P. Oldroyd and G. G. Malliaras, Acta Biomater., 2022, 139, 65.
305. B. Xu, et al. , Adv. Mater., 2016, 28, 4462.
306. D.-H. Kim, et al. , Science, 2011, 333, 838.
307. J. J. Norton, et al. , Proc. Natl. Acad. Sci., 2015, 112, 3920.
308. S. P. Lacour, D. Chan, S. Wagner, T. Li and Z. Suo, Appl. Phys. Lett., 2006, 88, 204103.
309. Y. Zhao, et al. , Science, 2022, 378, 1222.
310. X. Yu, et al. , Nature, 2019, 575, 473.
311. A. Fellner, A. Heshmat, P. Werginz and F. Rattay, J. Neural Eng., 2022, 19, 022001.
312. Y. S. Choi, et al. , Science, 2022, 376, 1006.
313. T. D. Kozai, et al. , Biomaterials, 2015, 37, 25.
314. J. Park, et al. , Adv. Funct. Mater., 2024, 2313728.
315. H. Choi, et al. , Nat. Electron., 2023, 6, 779.
316. H. Wu, et al. , Adv. Sci., 2021, 8, 2001938.
317. J. F. Kurniawan, et al. , Adv. Mater. Technol., 2021, 6, 2001229.
318. L. Tian, et al. , Nat. Biomed. Eng., 2019, 3, 194.
319. L. Sun, et al. , Chem. Eng. J., 2024, 481, 148096.
320. F. Ershad, et al. , Nat. Commun., 2020, 11, 3823.
321. D. W. Kim, et al. , Adv. Mater., 2022, 34, 2105338.
322. M. T. Flavin, et al. , Proc. Natl. Acad. Sci. U. S. A., 2022, 119, e2117764119.
323. B. Zhang, et al. , Nature, 2024, 628, 84.
324. Q. Zhang, et al. , ACS Nano, 2023, 17, 16798.
325. H. Lei and D. Fan, Chem. Eng. J., 2021, 421, 129578.
326. R. Horne, et al. , Acta Biomater., 2023, 166, 212.
327. L.-F. Wang, J.-Q. Liu, B. Yang, X. Chen and C.-S. Yang, Microsyst. Technol., 2015, 21, 1093.
328. C. Yuan, et al. , Adv. Mater. Interfaces, 2022, 9, 2200532.
329. Y. Wu, et al. , Adv. Funct. Mater., 2023, 33, 2210562.
330. S. Li, Y. Cong and J. Fu, J. Mater. Chem. B, 2021, 9, 4423.
331. P. Rao, et al. , Adv. Mater., 2018, 30, 1801884.
332. H. Min, et al. , ACS Appl. Mater. Interfaces, 2020, 12, 14425.
333. Y. H. Ju, et al. , ACS Appl. Mater. Interfaces, 2020, 12, 40794.
334. H. Wu, et al. , Sci. Transl. Med., 2023, 15, eabq1634.
335. Z. Liu and F. Yan, Adv. Sci., 2022, 9, 2200264.
336. K. R. Jinkins, et al. , Sci. Adv., 2022, 8, eabo0537.
337. L. Xue, et al. , Nano Lett., 2013, 13, 5541.
338. Y. Liu, et al. , Adv. Mater., 2022, 34, 2108820.
339. K. Imato, K. Momota, N. Kaneda, I. Imae and Y. Ooyama, Chem. Mater., 2022, 34, 8289.
340. Y. Lu, et al. , PLoS One, 2020, 15, e0236176.
341. H. Plenk Jr, Artif. Organs, 2011, 35, 237.
342. G. Schiavone, et al. , Neuron, 2020, 108, 238.
343. J. T. Mortimer and N. Bhadra, Neuromodulation, Elsevier, 2018, p. 71.
344. S. M. Wellman, et al. , Adv. Funct. Mater., 2018, 28, 1701269.
345. M. F. Ulum, W. Caesarendra, R. Alavi and H. Hermawan, Coatings, 2019, 9, 282.
346. S. Wang, et al. , Int. J. Biol. Macromol., 2024, 256, 128496.
347. K. Kang, et al. , Nat. Commun., 2024, 15, 10.
348. M. Santini and F. de Seta, Italian multicenter study croup on low output stimulation, Pacing Clin. Electrophysiol., 1993, 16, 722.
349. D. Chan, et al. , Adv. Mater., 2022, 34, 2109764.
350. Z. Huang and H. Ghasemi, Adv. Colloid Interface Sci., 2020, 284, 102264.
351. N. Erathodiyil, H.-M. Chan, H. Wu and J. Y. Ying, Mater. Today, 2020, 38, 84.
352. Y. Zhang, et al. , Chem. Rev., 2023, 123, 11722.
353. K. J. Yu, et al. , Nat. Mater., 2016, 15, 782.
354. K. J. Yu, et al. , Nat. Mater., 2016, 15, 782.
355. E. Vey, et al. , Polym. Degrad. Stab., 2008, 93, 1869.
356. Y. Yang, et al. , Nano-Micro Lett., 2022, 14, 161.
357. Y. Ding, et al. , Chem. Rev., 2024, 124, 1535.
358. X. Ma, et al. , Sci. Adv., 2023, 9, eadj2763.
359. S. Harimurti, et al. , Mater. Today, 2024, 74, 94.
360. Q. Zhang, et al. , Drug Discovery Today, 2017, 22, 1351.
361. Q. Tang, et al. , ACS Appl. Mater. Interfaces, 2023, 15, 17641.
362. S. Shi, et al. , Adv. Mater., 2023, 2306435.
363. Q. Huang and Z. Zheng, ACS Nano, 2022, 16, 15537.
364. B. Zhang, et al. , Nature, 2024, 1.
365. J. Deng, et al. , Nat. Mater., 2021, 20, 229.
366. L. Lan, et al. , Adv. Mater., 2024, 2401151.
367. B. Maity, H. Moorthy and T. Govindaraju, ACS Appl. Mater. Interfaces, 2023, 15, 49953.
368. S. Liu, et al. , Adv. Funct. Mater., 2021, 31, 2102225.
369. J.-W. Li, B.-S. Huang, C.-H. Chang and C.-W. Chiu, Adv. Compos. Hybrid Mater., 2023, 6, 231.
370. H. Zheng, et al. , Nat. Commun., 2019, 10, 2790.
371. Q. Zhang, H.-P. Bei, M. Zhao, Z. Dong and X. Zhao, Biomaterials, 2022, 286, 121566.
372. Y. Yang, et al. , Biomaterials, 2020, 263, 120378.
373. A. S. Rowlands and J. Cooper-White, J. Biomater., 2008, 29, 4510.
374. G. Shi, Z. Zhang and M. Rouabhia, Biomaterials, 2008, 29, 3792.
375. Y. Jiang, et al. , Nat. Biotechnol., 2023, 41, 652.
376. X. Zhang, et al. , Sci. Adv., 2023, 9, eadh1415.
377. G. Yao, et al. , Proc. Natl. Acad. Sci. U. S. A., 2021, 118, e2100772118.
378. Y. S. Choi, et al. , Nat. Commun., 2020, 11, 5990.
379. E. Shirzaei Sani, et al. , Sci. Adv., 2023, 9, eadf7388.
380. S. A. Eming, P. Martin and M. Tomic-Canic, Sci. Transl. Med., 2014, 6, 265sr6.
381. R. Luo, J. Dai, J. Zhang and Z. Li, Adv. Healthcare Mater., 2021, 10, 2100557.
382. M. A. Fonder, et al. , J. Am. Acad. Dermatol., 2008, 58, 185.
383. R. F. Diegelmann and M. C. Evans, Front. Biosci., 2004, 9, 283.
384. G. F. El Fawal, M. M. Abu-Serie, M. A. Hassan and M. S. Elnouby, Int. J. Biol. Macromol., 2018, 111, 649.
385. S. Bowers and E. Franco, Am. Fam. Physician, 2020, 101, 159.
386. J. H. Lee, W.-Y. Jeon, H.-H. Kim, E.-J. Lee and H.-W. Kim, Biomaterials, 2015, 53, 358.
387. Y. J. Cheah, M. R. Buyong and M. H. Mohd Yunus, Wound Healing with Electrical Stimulation Technologies: A Review, Polymers, 2021, 13, 3790.
388. I. S. Foulds and A. T. Barker, Br. J. Dermatol., 1983, 109, 515.
389. G. Talebi, et al., in 2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 3516.
390. M. Zhao, Semin. Cell Dev. Biol., 2009, 20, 674.
391. E. B. Nguyen, J. Wishner and K. Slowinska, J. Electroanal. Chem., 2018, 812, 265.
392. S.-H. Jeong, et al. , Nano Energy, 2021, 79, 105463.
393. C. Wang, et al. , Biomaterials, 2022, 285, 121479.
394. J. Cheng, et al. , Adv. Funct. Mater., 2022, 32, 2200444.
395. X. Huang, et al. , Biosens. Bioelectron., 2019, 124–125, 40.
396. X. Huang, et al. , Bio-Des. Manuf., 2022, 5, 201.
397. Z. L. Wang, Mater. Today, 2017, 20, 74.
398. W. Deng, et al. , Nano Energy, 2022, 91, 106656.
399. X. Xiao, et al. , Adv. Healthcare Mater., 2021, 10, 2100975.
400. W. Wang, et al. , ACS Nano, 2023, 17, 9793.
401. M. Pan, et al. , iScience, 2020, 23, 101682.
402. X. Wang and J. Liu, Recent Advancements in Liquid Metal Flexible Printed Electronics: Properties, Technologies, and Applications, Micromachines, 2016, 7, 206.
403. G. Li, et al. , Nat. Electron., 2023, 6, 154.
404. K. F. Cutting, Br. J. Community Nurs., 2003, 8, S4.
405. G. Power, Z. Moore and T. O'Connor, J. Wound Care, 2017, 26, 381.
406. J. Li, et al. , Bioeng. Transl. Med., 2023, 8, e10445.
407. R. Dong, P. X. Ma and B. Guo, Biomaterials, 2020, 229, 119584.
408. Z. Chen, B. Li, R.-Z. Zhan, L. Rao and N. Bursac, Sci. Adv., 2021, 7, eabd9502.
409. T. Dvir, et al. , Nat. Nanotechnol., 2011, 6, 720.
410. H. Sun, et al. , Acta Biomater., 2017, 48, 88.
411. S. Derhambakhsh, et al. , Tissue Cell, 2023, 81, 101996.
412. R. Langer and J. P. Vacanti, Science, 1993, 260, 920.
413. I. Apsite, et al. , Biofabrication, 2020, 12, 015016.
414. J. M. Razal, et al. , Adv. Funct. Mater., 2009, 19, 3381.
415. T. Jandova, Life, 2020, 10, 184.
416. B.-T. Zhang, et al. , BMC Cell Biol., 2010, 11, 87.
417. I. Jun, et al. , Bioact. Mater., 2022, 11, 118.
418. M. F. Ahmad and A. H. Hasbullah, Int. J. Sport Health Sci., 2015, 9, 864.
419. Y. Wang, et al. , Regener. Biomater., 2021, 8, rbab035.
420. W. Kim, et al. , Adv. Funct. Mater., 2021, 31, 2105170.
421. A. Khodabukus, et al. , Biomaterials, 2019, 198, 259.
422. A. Gelmi, M. K. Ljunggren, M. Rafat and E. W. H. Jager, J. Mater. Chem. B, 2014, 2, 3860.
423. H. Jo, et al. , Acta Biomater., 2017, 48, 100.
424. J. Zhou, et al. , Theranostics, 2018, 8, 3317.
425. S. Ribeiro, et al. , ACS Appl. Bio Mater., 2020, 3, 4239.
426. B. S. Spearman, et al. , Acta Biomater., 2015, 28, 109.
427. J. M. Khalifeh, et al. , IEEE Rev. Biomed. Eng., 2018, 11, 217.
428. Y. Yang, et al. , Small, 2021, 17, 2006598.
429. Z. Chen, et al. , Bioact. Mater., 2021, 6, 589.
430. F. Lin, et al. , Sci. Adv., 2024, 10, eadl3063.
431. F. Zhu, et al. , Tissue Eng., Part B, 2022, 29, 217.
432. M. Oishi and S. T. Onesti, Neurosurgery, 2000, 47, 1041.
433. N. Kahanovitz, Spine J., 2002, 2, 145.
434. J. B. Haddad, A. G. Obolensky and P. Shinnick, J. Altern. Complementary Med., 2007, 13, 485.
435. M. Griffin and A. Bayat, Eplasty, 2011, 11, e34.
436. J. Tian, et al. , Nano Energy, 2019, 59, 705.
437. C. Yu, et al. , Biomaterials, 2023, 301, 122266.
438. L. Cui, et al. , Biomaterials, 2020, 230, 119617.
439. R. Li, et al. , Cell Biochem. Biophys., 2014, 68, 449.
440. A. Höke, Nat. Clin. Pract. Neurol., 2006, 2, 448.
441. S. Javeed, A. H. Faraji, C. Dy, W. Z. Ray and M. R. MacEwan, Interdiscip. Neurosurg., 2021, 24, 101117.
442. A. Höke and T. Brushart, Exp. Neurol., 2010, 223, 1.
443. A. Pestronk, D. B. Drachman and J. W. Griffin, Exp. Neurol., 1980, 70, 65.
444. H. J. SEDDON, Brain, 1943, 66, 237.
445. T. Scholz, et al. , J. Reconstr. Microsurg., 2009, 25, 339.
446. S. Y. Fu and T. Gordon, J. Neurosci., 1995, 15, 3886.
447. J. Kimura, R. L. Rodnitzky and S.-H. Okawara, Neurology, 1975, 25, 989.
448. M. P. Willand, M.-A. Nguyen, G. H. Borschel and T. Gordon, Neurorehabilitation Neural Repair, 2015, 30, 490.
449. F. Jin, et al. , Adv. Mater., 2021, 33, 2104175.
450. L. Li, et al. , Adv. Mater., 2023, 35, 2302997.
451. B. N. Jensen, et al. , Npj Flex. Electron., 2023, 7, 34.
452. P. Wu, et al. , Nano Energy, 2022, 102, 107707.
453. L. Wang, et al. , Sci. Adv., 2020, 6, eabc6686.
454. W. Daly, L. Yao, D. Zeugolis, A. Windebank and A. Pandit, J. R. Soc., Interface, 2012, 9, 202.
455. G. Lundborg, J. Hand Surgery, 2000, 25, 391.
456. T. S. Pinho, C. B. Cunha, S. Lanceros-Méndez and A. J. Salgado, ACS Appl. Bio Mater.s, 2021, 4, 6604.
457. D. Seong, et al. , Adv. Mater., 2024, 36, 2307810.
458. E. Udina, et al. , Exp. Neurol., 2008, 210, 238.
459. J. Huang, Z. Ye, X. Hu, L. Lu and Z. Luo, Glia, 2010, 58, 622.
460. T. Gordon, E. Udina, V. M. K. Verge and E. I. P. de Chaves, Motor Control, 2009, 13, 412.
461. C. Aglah, T. Gordon and E. I. Posse de Chaves, Neuropharmacology, 2008, 55, 8.
462. B. Zorko, J. Rozman and A. Seliškar, Acta Vet. Hung., 2000, 48, 99.
463. E. M. Foecking, et al. , J. Rehabilitation Res. Dev., 2012, 49, 451.
464. T. M. Brushart, et al. , J. Neurosci., 2002, 22, 6631.
465. N. M. Geremia, T. Gordon, T. M. Brushart, A. A. Al-Majed and V. M. K. Verge, Exp. Neurol., 2007, 205, 347.
466. J. Huang, et al. , Neurorehabilitation Neural Repair, 2010, 24, 736.
467. S. Hasiba-Pappas, et al. , J. Pers. Med., 2023, 13, 414.
468. M. Rahman, T. Mahady Dip, R. Padhye and S. Houshyar, J. Biomed. Mater. Res., Part A, 2023, 111, 1916.
469. Z. Liu, et al. , BioMed Res. Int., 2020, 2020, 4794982.
470. N. Li, et al. , Sci. Rep., 2013, 3, 1604.
471. S. Lu, et al. , Small Methods, 2023, 7, 2200883.
472. R. L. Williams and P. J. Doherty, J. Mater. Sci.: Mater. Med., 1994, 5, 429.
473. H. Durgam, et al. , J. Biomater. Sci., Polym. Ed., 2010, 21, 1265.
474. J. Song, et al. , Front. Mol. Neurosci., 2016, 9, 117.
475. L. Ghasemi-Mobarakeh, M. P. Prabhakaran, M. Morshed, M. H. Nasr-Esfahani and S. Ramakrishna, Tissue Eng., Part A, 2009, 15, 3605.
476. D. K. Cullen, A. R. Patel, J. F. Doorish, D. H. Smith and B. J. Pfister, J. Neural Eng., 2008, 5, 374.
477. M. R. Abidian, J. M. Corey, D. R. Kipke and D. C. Martin, Small, 2010, 6, 421.
478. B. Weng, et al. , Adv. Mater. Interfaces, 2015, 2, 1500059.
479. Y. Qian, et al. , Small, 2020, 16, 2000796.
480. R. Mao, et al. , Nano Energy, 2022, 98, 107322.
481. M. Mohseni and A. Ramazani Saadatabadi, Appl. Nanosci., 2021, 11, 2199.
482. M. Zhou, et al. , Adv. Funct. Mater., 2022, 32, 2200269.
483. C. L. Schmidt and P. M. Skarstad, J. Power Sources, 2001, 97–98, 742.
484. Y. Sun, et al. , Adv. Healthcare Mater., 2019, 8, 1900127.
485. J.-H. Fang, et al. , NPG Asia Mater., 2020, 12, 61.
486. J. Koo, et al. , Nat. Med., 2018, 24, 1830.
487. C. M. Boutry, et al. , Philos. Trans. R. Soc., A, 2012, 370, 2418.
488. L. Wang, et al. , Sci. Adv., 2020, 6, eabc6686.
489. H. Joo, et al. , Sci. Adv., 2021, 7, eabd4639.
490. S. M. Mirvakili and R. Langer, Nat. Electron., 2021, 4, 464.
491. X. Jiang, et al. , Small Methods, 2024, 8, 2301068.
492. S. Zhao, A. S. Mehta and M. Zhao, Cell. Mol. Life Sci., 2020, 77, 2681.
493. A. J. Simon, et al. , Vaccine, 2008, 26, 5202.
494. X. Huang, et al. , InfoMat, 2023, 5, e12388.
495. Y. Ogawa, et al. , Adv. Healthcare Mater., 2015, 4, 506.
496. C. Wu, et al. , Adv. Funct. Mater., 2020, 30, 1907378.
497. S. Kusama, et al. , Nat. Commun., 2021, 12, 658.
498. Y. Wang, et al. , Nat. Commun., 2024, 15, 511.
499. X. Li, et al. , Adv. Sci., 2021, 8, 2100827.
500. G. Xu, et al. , Adv. Funct. Mater., 2021, 31, 2100852.
501. Y. Yang, et al. , Adv. Funct. Mater., 2021, 31, 2104092.
502. Y. Chen, Q. An, K. Teng, Y. Zhang and Y. Zhao, Eur. Polym. J., 2022, 170, 111164.
503. S. Bian, B. Zhu, G. Rong and M. Sawan, J. Pharm. Anal., 2021, 11, 1.
504. N. F. Nadia Ahmad, N. N. Nik Ghazali and Y. H. Wong, Biosens. Bioelectron., 2021, 189, 113384.
505. L. Chang, et al. , Lab Chip, 2016, 16, 4047.
506. Z. Dong, et al. , Nano Lett., 2021, 21, 4878.
507. X. Xu, et al. , Adv. Funct. Mater., 2024, 34, 2311144.
508. D. Xia, et al. , Proc. Natl. Acad. Sci. U. S. A., 2021, 118, e2110817118.
509. Z. Wei, et al. , Lab Chip, 2014, 14, 4093.
510. M. Cho, et al. , Nat. Commun., 2024, 15, 2000.
511. Y. Chen, et al. , Biomaterials, 2021, 268, 120559.
512. M. Dümpelmann, J, Neural Eng., 2019, 16, 041001.
513. A. S. Widge, et al. , Exp. Neurol., 2017, 287, 461.
514. Y. Cho, S. Park, J. Lee and K. J. Yu, Adv. Mater., 2021, 33, 2005786.
515. E. McGlynn, et al. , Adv. Sci., 2021, 8, 2002693.
516. H. Li, J. Wang and Y. Fang, Nanoscale Adv., 2020, 2, 3095.
517. R. Sitaram, et al. , Nat. Rev. Neurosci., 2017, 18, 86.
518. P. Li, et al. , Nature, 2024, 626, 990.
519. M. Vöröslakos, et al. , Nat. Commun., 2018, 9, 483.
520. T. M. Herrington, J. J. Cheng and E. N. Eskandar, J. Neurophysiol., 2015, 115, 19.
521. K. Ashkan, P. Rogers, H. Bergman and I. Ughratdar, Nat. Rev. Neurol., 2017, 13, 548.
522. J. O’Doherty, M. Lebedev, T. Hanson, N. Fitzsimmons and M. Nicolelis, Front. Integr. Neurosci., 2009, 3, 803.
523. M. A. Schwemmer, et al. , Nat. Med., 2018, 24, 1669.
524. R. van den Brand, et al. , Science, 2012, 336, 1182.
525. M. P. Powell, et al. , Nat. Med., 2023, 29, 689.
526. Y. Yang, Y. Tang, H. Qin and J. Xu, Spinal Cord, 2022, 60, 375.
527. F. B. Wagner, et al. , Nature, 2018, 563, 65.
528. D. J. Chew, et al. , Sci. Transl. Med., 2013, 5, 210ra155.
529. M. S. George, et al. , Biol. Psychiatry, 2000, 47, 287.
530. J. Yang and J. H. Phi, J. Korean Neurosurg. Soc., 2019, 62, 344.
531. Y. Cho, J. Park, C. Lee and S. Lee, Bioelectron. Med., 2020, 6, 23.
532. D. A. Groves and V. J. Brown, Neurosci. Biobehav. Rev., 2005, 29, 493.
533. R. L. Johnson and C. G. Wilson, J. Inflammation Res., 2018, 203.
534. J. L. Collinger, et al. , Lancet, 2013, 381, 557.
535. L. R. Hochberg, et al. , Nature, 2012, 485, 372.
536. Brain Dynamics: An Introduction to Models and Simualtions, ed. H. Haken, Springer, Berlin Heidelberg, 2008, pp. 9.
537. J. C. Fiala and K. M. Harris, Dendrites, 1999, 2, 1.
538. G. J. Augustine, et al., in Neuroscience The Somatic Sensory System, ed. D. Purves, et al., Sinauer Associates Inc, 2004, ch. 8, p. 189.
539. C. T. Tsui, P. Lal, K. V. R. Fox, M. A. Churchward and K. G. Todd, BMC Biomed. Eng., 2022, 4, 7.
540. Y. David, J. H. Wittig Jr, J. Samantha, K. I. Sara and A. Z. Kareem, J. Neurosci. Res., 2023, 43, 4448.
541. D. R. McNeal, IEEE Trans. Biomed. Eng., 1976, 23, 329.
542. G. Weiss, Arch. Ital. Biol., 1901, 35, 413.
543. T.-i Kim, et al. , Science, 2013, 340, 211.
544. S. Kim, et al. , Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 15202.
545. T. E. Schlaepfer, et al. , Neuropsychopharmacology, 2008, 33, 368.
546. A. M. Lozano, et al. , Nat. Rev. Neurol., 2019, 15, 148.
547. P. Hickey and M. Stacy, Front. Neurosci., 2016, 10, 173.
548. J. Kim, T. Wichmann, O. T. Inan and S. P. Deweerth, IEEE J. Transl. Eng. Health Med., 2020, 8, 1.
549. O. G. Sani, et al. , Nat. Biotechnol., 2018, 36, 954.
550. T. M. Herrington, J. J. Cheng and E. N. Eskandar, J. Neurophysiol., 2016, 115, 19.
551. K. J. van Dijk, et al. , J. Neural Eng., 2015, 12, 046003.
552. R. Lycke, et al. , Cell Rep., 2023, 42, 112554.
553. T.-M. Fu, et al. , Nat. Methods, 2016, 13, 875.
554. S. Zhao, et al. , Nat. Commun., 2020, 11, 1788.
555. J. K. Mai, M. Majtanik and G. Paxinos, Atlas of the human brain, Academic Press, 2015.
556. L. Timmermann, et al. , Lancet Neurol., 2015, 14, 693.
557. M. S. Troche, et al. , Dysphagia, 2014, 29, 425.
558. S. Tisch and P. Limousin, Exp. Brain Res., 2020, 238, 1645.
559. T. Tsuboi, et al. , Neurology, 2020, 94, e1073.
560. L. Mar-Barrutia, et al. , Brain Stimul., 2022, 15, 1128.
561. T. J. Foutz and M. Wong, Biomed. J., 2022, 45, 27.
562. C. Villard, et al. , Cortex, 2023, 164, 1.
563. M. M. Shanechi, Nat. Neurosci., 2019, 22, 1554.
564. K. Lee, et al. , Nat. Commun., 2024, 15, 218.
565. J. B. Shute, et al. , NeuroImage: Clin., 2016, 12, 165.
566. M. Z. Zarzycki and I. Domitrz, Acta Neuropsychiatrica, 2020, 32, 57.
567. M. Fabbri, et al. , Parkinsonism Relat. Disord., 2021, 87, 130.
568. G. D. Pal, et al. , Mov. Disord, 2023, 38, 2155.
569. H. Cagnan, T. Denison, C. McIntyre and P. Brown, Nat. Biotechnol., 2019, 37, 1024.
570. J. Evers and M. Lowery, Oper. Neurosurgery, 2021, 20, 131.
571. J. Evers, et al. , J. Neural Eng., 2022, 19, 046004.
572. S. Park, et al. , Nat. Neurosci., 2017, 20, 612.
573. J. Wang, et al. , Adv. Mater., 2023, 35, 2211012.
574. C. Xie, et al. , Nat. Mater., 2015, 14, 1286.
575. S. Guan, et al. , Sci. Adv., 2019, 5, eaav2842.
576. G. Hong, et al. , Nano Lett., 2015, 15, 6979.
577. L. Tian, et al. , Nat. Biomed. Eng., 2019, 3, 194.
578. K. Huang, et al. , Adv. Mater., 2024, 2313752.
579. R. Green and M. R. Abidian, Adv. Mater., 2015, 27, 7620.
580. B. D. Greenberg, et al. , Neuropsychopharmacology, 2006, 31, 2384.
581. S. Thobois, et al. , J. Neurol., 2002, 249, 529.
582. J. Vesper, F. Klostermann, F. Stockhammer, T. Funk and M. Brock, Surg. Neurol., 2002, 57, 306.
583. P. Doshi and P. Bhargava, Neurol. India, 2008, 56, 474.
584. T. Soulas, et al. , J. Neurol., Neurosurg. Psychiatry, 2008, 79, 952.
585. P. Krack, et al. , N. Engl. J. Med., 2003, 349, 1925.
586. L. Wojtecki, et al. , Mov. Disord, 2007, 22, 1314.
587. L. Ackermans, et al. , Brain, 2011, 134, 832.
588. M. Mantione, M. Figee and D. Denys, Front. Behav. Neurosci., 2014, 8, 152.
589. M. Mantione, et al. , J. Psychiatry Neurosci., 2015, 40, 378.
590. D. Denys, et al. , Arch. Gen. Psychiatry, 2010, 67, 1061.
591. T. A. Dembek, et al. , Mov. Disord, 2017, 32, 1380.
592. J. Wu, et al. , Nature, 2024, 630, 360.
593. W. H. Lee, S. H. Lisanby, A. F. Laine and A. V. Peterchev, IEEE Trans. Biomed. Eng., 2015, 62, 2095.
594. A. Liu, et al. , Nat. Commun., 2018, 9, 5092.
595. C. Hughes, A. Herrera, R. Gaunt and J. Collinger in Handbook of Clinical Neurology, ed. N. F. Ramsey and J. D. R. Millán, Elsevier, 2020, vol. 168, pp. 163.
596. A. Antal, et al. , Brain Stimul., 2008, 1, 97.
597. M. R. Krause, et al. , Curr. Biol., 2017, 27, 3086.
598. A. Antal and W. Paulus, Front. Hum. Neurosci., 2013, 7, 317.
599. S.-W. Park, et al. , Small, 2018, 14, 1801732.
600. I. Uguz and K. L. Shepard, Sci. Adv., 2022, 8, eabq6354.
601. S. N. Flesher, et al. , Sci. Transl. Med., 2016, 8, 361ra141.
602. N. L. Opie, et al. , Nat. Biomed. Eng., 2018, 2, 907.
603. J. Lee, et al. , Nat. Electron., 2021, 4, 604.
604. S. V. Hiremath, et al. , PLoS One, 2017, 12, e0176020.
605. R. S. Johansson and J. R. Flanagan, Nat. Rev. Neurosci., 2009, 10, 345.
606. S. N. Flesher, et al. , Science, 2021, 372, 831.
607. F.-G. Zeng, P. Tran, M. Richardson, S. Sun and Y. Xu, Sci. Rep., 2019, 9, 15247.
608. W. H. Bosking, M. S. Beauchamp and D. Yoshor, Annu. Rev. Vis. Sci., 2017, 3, 141.
609. A. J. Lowery, et al., in 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 1041.
610. M. R. Bower, et al. , J. Neurosci. Methods, 2013, 214, 21.
611. P. Stoeter, L. Dieterle, A. Meyer and N. Prey, Am. J. Neuroradiology, 1995, 16, 1214.
612. T. J. Oxley, et al. , Nat. Biotechnol., 2016, 34, 320.
613. N. L. Opie, et al. , J. Neural. Eng., 2016, 13, 046020.
614. A. Burton, et al. , Microsyst. Nanoeng., 2021, 7, 62.
615. C. Ethier, E. R. Oby, M. J. Bauman and L. E. Miller, Nature, 2012, 485, 368.
616. M. A. Lebedev and M. A. L. Nicolelis, Physiol. Rev., 2017, 97, 767.
617. K. A. Moxon and G. Foffani, Neuron, 2015, 86, 55.
618. C. E. Bouton, et al. , Nature, 2016, 533, 247.
619. W. Dong, et al. , Int. J. Intell. Rob. Appl., 2018, 2, 313.
620. U. Chaudhary, N. Birbaumer and A. Ramos-Murguialday, Nat. Rev. Neurol., 2016, 12, 513.
621. M. Capogrosso, et al. , Nature, 2016, 539, 284.
622. A. D. Sdrulla, Y. Guan and S. N. Raja, Pain Pract., 2018, 18, 1048.
623. J. P. Hunter and P. Ashby, J. Physiol., 1994, 474, 407.
624. R. B. North, D. H. Kidd, L. Petrucci and M. J. Dorsi, Neurosurgery, 2005, 57, 990.
625. L. Kapural, et al. , Anesthesiology, 2015, 123, 851.
626. W. K. Sin and B. Coburn, Med. Biol. Eng. Comput., 1983, 21, 264.
627. J. Holsheimer, Neuromodulation: Technol. Neural Interface, 2002, 5, 25.
628. J. Ladenbauer, K. Minassian, U. S. Hofstoetter, M. R. Dimitrijevic and F. Rattay, IEEE Trans. Neural Syst. Rehabilitation Eng., 2010, 18, 637.
629. M. Capogrosso, et al. , Nat. Protoc., 2018, 13, 2031.
630. N. Wenger, et al. , Nat. Med., 2016, 22, 138.
631. D. A. Stidd, S. Rivero and M. E. Weinand, J. Pain Res., 2014, 465.
632. I. R. Minev, et al. , Science, 2015, 347, 159.
633. B. J. Woodington, et al. , Sci. Adv., 2021, 7, eabg7833.
634. B. Linderoth and R. D. Foreman, Neuromodulation: Technol. Neural Interface, 1999, 2, 150.
635. K. Minassian and U. S. Hofstoetter, CNS Neurosci. Ther., 2016, 22, 262.
636. C. Kathe, et al. , Nature, 2022, 611, 540.
637. B. J. Woodington, et al. , Sci. Adv., 2024, 10, eadl1230.
638. R. Melzack and P. D. Wall, Science, 1965, 150, 971.
639. C. Nezhat, R. Falik, S. McKinney and L. P. King, Nat. Rev. Urol., 2017, 14, 359.
640. M. Brazzelli, A. Murray and C. Fraser, J. Urology, 2006, 175, 835.
641. U. Barroso and P. Lordêlo, Nat. Rev. Urol., 2011, 8, 402.
642. M. J. Hilz, Auton. Neurosci., 2022, 243, 103038.
643. D. J. Lanska, Neurology, 2002, 58, 452.
644. S. E. Krahl, Surg. Neurol. Int., 2012, 3, S47.
645. B. M. Uthman, Arch. Med. Res., 2000, 31, 300.
646. F. A. Koopman, et al. , Proc. Natl. Acad. Sci., 2016, 113, 8284.
647. A. Kirchner, et al. , J. Clin. Neurophysiol., 2004, 21, 418.
648. W. Sperling, et al. , Pharmacopsychiatry, 2011, 44, 67.
649. R. L. Johnson and C. G. Wilson, J. Inflammation Res., 2018, 11, 203.
650. M. O. Jakob, S. Murugan and C. S. N. Klose, Front. Immunol., 2020, 11, 511994.
651. Z. U. A. Asad, et al. , Bioelectron. Med., 2020, 3, 51.
652. S. C. Payne, et al. , Front. Neurosci., 2019, 13, 418.
653. A. R. Murray, L. Atkinson, M. K. Mahadi, S. A. Deuchars and J. Deuchars, Auton. Neurosci., 2016, 199, 48.
654. W. He, et al. , J. Evidence-Based Complementary Altern. Med., 2012, 2012, 786839.
655. D. C. Baumgart and W. J. Sandborn, Lancet, 2007, 369, 1641.
656. A. N. Ananthakrishnan, Nat. Rev. Gastroenterol. Hepatol., 2015, 12, 205.
657. J. Meregnani, et al. , Auton. Neurosci., 2011, 160, 82.
658. P. Sun, et al. , PLoS One, 2013, 8, e69424.
659. O. Jacobowitz, Upper Airway Stimulation in Obstructive Sleep Apnea: Best Practices in Evaluation and Surgical Management, Springer, 2022, p. 123.
660. S. Lee, et al. , Adv. Sci., 2017, 4, 1700149.
661. S. Lee, et al. , Sens. Actuators, B, 2017, 242, 1165.
662. Y. Liu, et al. , Nat. Biotechnol., 2020, 38, 1031.
663. Y. J. Hong, H. Jeong, K. W. Cho, N. Lu and D.-H. Kim, Adv. Funct. Mater., 2019, 29, 1808247.
664. J. Ausra, et al. , Sci. Adv., 2022, 8, eabq7469.
665. P. Gutruf, et al. , Nat. Commun., 2019, 10, 5742.
666. Y. S. Choi, et al. , Science, 2022, 376, 1006.
667. N. Li, et al. , ACS Nano, 2019, 13, 2822.
668. H. Ryu, et al. , Nat. Commun., 2021, 12, 4374.
669. E. Elmistekawy, Asian Cardiovasc. Thorac. Ann., 2019, 27, 341.
670. P. D. Nido and B. S. Goldman, J. Cardiac Surgery, 1989, 4, 99.
671. H. K. Makadia and S. J. Siegel, Polymers, 2011, 3, 1377.
672. F. Witte, Acta Biomater., 2010, 6, 1680.
673. V. S. Mallela, V. Ilankumaran and N. S. Rao, Indian Pacing Electrophysiol. J., 2004, 4, 201.
674. S. Azimi, et al. , Nano Energy, 2021, 83, 105781.
675. L. Dong, et al. , Nano Energy, 2019, 66, 104085.
676. Q. Zheng, et al. , Adv. Mater., 2014, 26, 5851.
677. H. Lyu, et al. , Sci. Rep., 2020, 10, 2067.
678. S. N. Raja, et al. , Pain, 2020, 161, 1976.
679. P. Seth, R. A. Rudd, R. K. Noonan and T. M. Haegerich, Am. J. Public Health, 2018, 108, 500.
680. I. Jones and M. I. Johnson, Contin. Educ. Anaesth. Crit. Care Pain, 2009, 9, 130.
681. K. A. Sluka and D. Walsh, J. Pain, 2003, 4, 109.
682. M. Johnson, Rev. Pain, 2007, 1, 7.
683. A. de Sire, et al., Percutaneous Electrical Nerve Stimulation (PENS) as a Rehabilitation Approach for Reducing Mixed Chronic Pain in Patients with Musculoskeletal Disorders, Appl. Sci., 2021, 11, 4257.
684. T. R. Deer, et al. , J. Pain Res., 2021, 14, 721.
685. A. M. Cobo, et al. , J. Microelectromech. Syst., 2019, 28, 36.
686. B. H. Sjölund and M. B. E. Eriksson, Brain Res., 1979, 173, 295.
687. A. de Sire, et al. , Appl. Sci., 2021, 11, 4257.
688. M. Rossi, et al. , Pain Physician, 2016, 19, E121.
689. T. R. Deer, et al. , Bioelectron. Med., 2020, 6, 9.
690. G. Plaza-Manzano, et al. , Eur. J. Pain, 2020, 24, 1023.
691. R. A. Gabriel and B. M. Ilfeld, Expert Rev. Med. Devices, 2021, 18, 145.
692. R. L. Rauck, et al. , Neuromodulation: Technol. Neural Interface, 2014, 17, 188.
693. L. Botelho, et al. , Front. Hum. Neurosci., 2018, 12, 388.
694. M. da Graca-Tarragó, et al. , Pain Med., 2016, 17, 877.
695. H. Beltran-Alacreu, et al. , Pain Med., 2022, 23, 1387.
696. T. Eggers, et al. , J. Neural Eng., 2021, 18, 046010.
697. N. Bhadra, T. L. Vrabec, N. Bhadra and K. L. Kilgore, Bioelectron. Med., 2018, 1, 39.
698. E. L. Foldes, D. M. Ackermann, N. Bhadra and K. L. Kilgore in 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 614.
699. M.-A. Thil, D. T. Duy, I. M. Colin and J. Delbeke, J. Neuroimmunol., 2007, 185, 103.
700. B. M. Ilfeld, et al. , Neuromodulation: Technol. Neural Interface, 2019, 22, 621.
701. S. Dosen, et al. , IEEE Trans. Neural Syst. Rehabilitation Eng., 2015, 23, 385.
702. M.-Z. Hao, et al. , J. NeuroEngineering Rehabilitation, 2017, 14, 75.
703. P. T. Lin, et al. , Mov. Disord. Off. J. Mov. Disord. Soc., 2018, 33, 1182.
704. J. A. Herron, et al. , IEEE Trans. Neural Syst. Rehabilitation Eng., 2017, 25, 2180.
705. A. Pascual-Valdunciel, et al. , J. NeuroEng. Rehabil., 2021, 18, 33.
706. P. T. Lin, et al. , Mov. Disord, 2018, 33, 1182.
707. M. Bange, et al. , Nat. Commun., 2024, 15, 3166.
708. L. Popović Maneski, et al. , Med. Biol. Eng. Comput., 2011, 49, 1187.
709. A. P. L. Bó, C. Azevedo-Coste, C. Geny, P. Poignet and C. Fattal, Artif. Organs, 2014, 38, 984.
710. F. Widjaja, C. Y. Shee, W. L. Au, P. Poignet and W. T. Ang in 2011 IEEE International Conference on Robotics and Automation. 3694.
711. A. Pascual-Valdunciel, et al. , IEEE Trans. Biomed. Eng., 2021, 68, 1768.
712. R. Pahwa, et al. , Neuromodulation: Technol. Neural Interface, 2019, 22, 537.
713. F. L. Xu, et al., in 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 1782.
714. J. Zhang, Hear. Res., 2013, 295, 38.
715. Y. Zhou, et al. , Sci. Rob., 2024, 9, eadi8666.
716. J. Park, et al. , Sci. Rob., 2024, 9, eadk6903.
717. R. V. Shannon in Encyclopedia of Computational Neuroscience, ed. D. Jaeger and R. Jung, Springer, New York, 2013, p. 1.
718. F. G. Zeng, S. Rebscher, W. Harrison, X. Sun and H. Feng, IEEE Rev. Biomed. Eng., 2008, 1, 115.
719. S. Irving, et al. , Hear. Res., 2013, 306, 37.
720. Y. N. Ertas, D. Ozpolat, S. N. Karasu and N. Ashammakhi, Micromachines, 2022, 13, 1081.
721. M. Chen, S. Min, C. Zhang, X. Hu and S. Li, Neuromodulation Technol. Neural Interface, 2022, 25, 1338.
722. D. Gupta, P. Mathur and P. Tewari, in Applications of Computing, Automation and Wireless Systems in Electrical Engineering, ed. S. Mishra, Y. R. Sood and A. Tomar, Springer, Singapore, 2019, p. 851.
723. D. Gupta, P. Mathur and P. Tewari in Applications of Computing, Automation and Wireless Systems in Electrical Engineering: Proceedings of MARC 2018. 851 (Springer).
724. A. N. Dalrymple, et al. , J. Neural Eng., 2020, 17, 036012.
725. J. Jang, et al. , Adv. Healthcare Mater., 2019, 8, 1900379.
726. P. M. Lewis, H. M. Ackland, A. J. Lowery and J. V. Rosenfeld, Brain Res., 2015, 1595, 51.
727. A. Sehic, et al. , Am. J. Pathol., 2016, 186, 2783.
728. Y. H.-L. Luo and L. da Cruz, Prog. Retinal Eye Res., 2016, 50, 89.
729. V. Gaillet, et al. , Nat. Biomed. Eng., 2020, 4, 181.
730. A. D. Cheok and K. Karunanayaka in Virtual Taste and Smell Technologies for Multisensory Internet and Virtual Reality, ed. A. D. Cheok and K. Karunanayaka, Springer International Publishing, 2018, p. 93.
731. A. Cheok and K. Karunanayaka, Virtual Taste and Smell Technologies for Multisensory Internet and Virtual Reality, Electric taste, 2018, ch. 3, vol. 49, pp. 49–68.
732. E. Bas, et al. , Hear. Res., 2016, 337, 12.
733. H. Jia, et al. , Ann. Otol., Rhinol., Laryngol., 2013, 122, 33.
734. E. Bas, C. T. Dinh, C. Garnham, M. Polak and T. R. Van de Water, Anat. Rec., 2012, 295, 1909.
735. J. N. Fayad, A. O. Makarem and F. H. Linthicum Jr, Otolaryngol.–Head Neck Surg., 2009, 141, 247.
736. T. Lenarz, et al. , Audiol. Neurotol., 2006, 11, 34.
737. G. Paasche, F. Bockel, C. Tasche, A. Lesinski-Schiedat and T. Lenarz, Otol. Neurotol., 2006, 27, 639.
738. G. Paasche, C. Tasche, T. Stöver, A. Lesinski-Schiedat and T. Lenarz, Otol. Neurotol., 2009, 30, 592.
739. Y. Luo, et al. , Drug Delivery, 2021, 28, 1673.
740. I. Linke, et al. , Phys. Status Solidi A, 2015, 212, 1210.
741. H. H. Lim and D. J. Anderson, J. Neurophysiol., 2006, 96, 975.
742. A. E. Hadjinicolaou, H. Meffin, M. I. Maturana, S. L. Cloherty and M. R. Ibbotson, Clin. Exp. Optom., 2015, 98, 395.
743. E. M. Maynard, Annu. Rev. Biomed. Eng., 2001, 3, 145.
744. T. Fujikado, et al. , Graefe's Arch. Clin. Exp. Ophthalmol., 2007, 245, 1411.
745. J. Xie, et al. , IEEE Trans. Biomed. Eng., 2011, 58, 1932.
746. L. B. Merabet, J. F. Rizzo, A. Amedi, D. C. Somers and A. Pascual-Leone, Nat. Rev. Neurosci., 2005, 6, 71.
747. K. Stingl, et al. , Vision Res., 2015, 111, 149.
748. C. Veraart, et al. , Brain Res., 1998, 813, 181.
749. M. E. Brelén, et al. , J. Neurosurgery, 2006, 104, 593.
750. C. Veraart, M.-C. Wanet-Defalque, B. Gérard, A. Vanlierde and J. Delbeke, Artif. Organs, 2003, 27, 996.
751. M. E. Brelén, F. Duret, B. Gérard, J. Delbeke and C. Veraart, J. Neural Eng., 2005, 2, S22.
752. Y. Yan, et al. , Invest Ophthalmol. Visual Sci., 2016, 57, 6327.
753. J. Sun, Y. Chen, X. Chai, Q. Ren and L. Li, Graefe's Arch. Clin. Exp. Ophthalmol., 2013, 251, 2545.
754. Y. Lu, et al. , J. Neural Eng., 2013, 10, 036022.
755. A. D. Cheok and K. Karunanayaka, Virtual taste and smell technologies for multisensory internet and virtual reality, Springer, 2018.
756. D. E. García-Díaz, H. U. Aguilar-Baturoni, R. Guevara-Aguilar and M. J. Wayner, Brain Res. Bull., 1984, 12, 529.
757. A. Maharjan, E. Wang, M. Peng and Y. O. Cakmak, Front. Neurosci., 2018, 12, 225.
758. T. Weiss, et al. , Cereb. Cortex, 2016, 26, 4180.
759. T. H. Musiolik in Neuromarketing in Business: Identifying Implicit Purchase Drivers and Leveraging them for Sales, ed. B. B. Briesemeister and W. K. Selmer, Springer, Fachmedien Wiesbaden, 2022, p. 1.
760. T. Ishimaru, et al. , Chem. Senses, 1997, 22, 77.
761. T. Ishimaru, T. Miwa, T. Shimada and M. Furukawa, Ann. Otol., Rhinol., Laryngol., 2002, 111, 518.
762. C. Yamamoto, Jpn. J. Physiol., 1961, 11, 545.
763. D. A. Stevens, et al. , Chem. Senses, 2008, 33, 405.
764. N. Ranasinghe and E. Y.-L. Do, ACM Trans. Multimedia Comput. Commun. Appl, 2016, 13, 5.
765. S. A. Ahmad, V. Singh and S. Singh, Int. J. Eng. Res. Technol., 2020, 9, 1385.
766. N. Ranasinghe, A. Cheok, R. Nakatsu and E. Y.-L. Do in Proceedings of the 2013 ACM international workshop on Immersive media experiences 29 (Association for Computing Machinery, Barcelona, Spain, 2013).
767. T. P. Hettinger and M. E. Frank, Brain Res. Bull., 2009, 80, 107.
768. Y. Aruga and T. Koike in Proceedings of the 6th Augmented Human International Conference 191 (Association for Computing Machinery, Singapore, Singapore, 2015).
769. H. T. Lawless, D. A. Stevens, K. W. Chapman and A. Kurtz, Chem. Senses, 2005, 30, 185.
770. K. Sakurai, K. Aoyama, M. Mizukami, T. Maeda and H. Ando in Proceedings of the 1st Workshop on Multi-sensorial Approaches to Human-Food Interaction Article 2 (Association for Computing Machinery, Tokyo, Japan, 2016).
771. K. Aoyama, et al. , Front. Psychol., 2017, 8, 2112.
772. H. P. Saal, B. P. Delhaye, B. C. Rayhaun and S. J. Bensmaia, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, E5693.
773. B. Xu, et al. , Adv. Mater., 2016, 28, 4462.
774. G. Valle, et al. , Neuron, 2018, 100, 37.
775. A. Akhtar, J. Sombeck, B. Boyce and T. Bretl, Sci. Rob., 2018, 3, eaap9770.
776. K. Yao, et al. , Nat. Mach. Intell., 2022, 4, 893.
777. L. E. Osborn, et al. , Sci. Rob., 2018, 3, eaat3818.
778. D. W. Tan, et al. , Sci. Transl. Med., 2014, 6, 257ra138.
779. M. Ortiz-Catalan, B. Håkansson and R. Brånemark, Sci. Transl. Med., 2014, 6, 257re6.
780. S. Raspopovic, et al. , Sci. Transl. Med., 2014, 6, 222ra19.
781. G. Gu, et al. , Nat. Biomed. Eng., 2023, 7, 589.
782. E. D’Anna, et al. , Sci. Rob., 2019, 4, eaau8892.
783. C. Wu, et al. , Adv. Funct. Mater., 2020, 30, 1907378.

---

Footnote

† These authors contributed equally.

---