High-sensitivity SpectroChip-integrated LFIA platform for rapid point-of-care quantification of cardiovascular biomarkers

Abstract

Timely and precise measurement of cardiovascular biomarkers is crucial for early diagnosis and effective management of heart disease. This study evaluates THE ONE InstantCare device, a point-of-care (POC) platform that integrates a SpectroChip technology with gold nanoparticles (AuNPs)-conjugated lateral flow immunoassays (LFIA). Using this platform, we quantified five clinically significant cardiovascular biomarkers: Troponin I (cTnI), Creatine Kinase Myocardial Band (CK-MB), N-terminal pro-B-type Natriuretic Peptide (NT-proBNP), D-dimer, and C-reactive Protein (CRP). Standard curves were generated from serial dilutions of known concentrations spiked into fetal bovine serum (FBS). Spectral data were averaged across three replicates and preprocessed using baseline correction and noise reduction techniques. Relative absorbance (ΔA) was calculated by subtracting the absorbance at 680 nm (baseline) from that at 530 nm (peak), minimizing baseline interference and enhancing quantitative accuracy. The platform achieved limits of detection (LOD) as low as 0.0124 ng mL⁻¹ (cTnI), 0.0342 ng mL⁻¹ (CK-MB), 0.1422 ng mL⁻¹ (NT-proBNP), 0.0213 ng L⁻¹ (D-dimer), and 0.0304 ng L⁻¹ (CRP) with R² values above 0.98 for all calibration curves. Clinical validation using 47 patient serum samples showed strong agreement with the Abbott Alinity ci-series analyzer (R² > 0.91, r > 0.95). Bland–Altman analysis confirmed consistency within 95% confidence intervals. These results highlight the potential of the proposed novel Point-of-Care (POC) system to deliver fast, reliable and affordable cardiovascular diagnostics in emergency care settings.

---

1. Introduction

Globally, cardiovascular diseases (CVDs) are the leading cause of mortality, accounting for 51.7% of all deaths in 2019, which is before the emergency of the COVID-19 pandemic.¹ The pandemic has exacerbated global cardiovascular health issues, significantly increasing the risk of long-term complications such as heart failure due to inflammation and myocardial injury.^2,3 Additionally, it has strained healthcare systems, disrupted routine cardiovascular care, reduced hospital admissions for acute coronary syndromes, delayed treatments, and increased mortality rates among those with pre-existing conditions.^3–6 These challenges highlight the urgent need for innovative diagnostic approaches, such as point-of-care (POC) testing, to effectively manage and accurately detect cardiac conditions.^7–9

Cardiovascular biomarkers are essential for early detection and effective management of conditions such as acute myocardial infarction (AMI). Troponin (cTnI) and Creatine Kinase-MB (CK-MB) are particularly valuable for their high specificity and sensitivity to myocardial damage.¹⁰ cTnI is known for its ability of detect even minor myocardial injuries,¹¹ and remains elevated for days, making it a reliable marker for diagnosis and ongoing monitoring.¹² Although CK-MB's specificity may be compromised by skeletal muscle injuries, it proves effective when used in conjunction with cTnI shortly after an event.¹³ N-terminal pro-B-type Natriuretic Peptide (NT-proBNP) is useful for assessing cardiac stress and predicting future events post-MI (Myocardial Infarction),¹⁴ while C-Reactive Protein (CRP) evaluates myocardial inflammation and potential MI risks.¹⁵D-Dimer, an indicator of thrombosis and overall vascular health.¹⁶ Integrating these biomarkers into clinical practice improves early detection and risk stratification, and informs treatment strategies by revealing the extent of myocardial damage and cardiac stress.^17–19 Despite their benefits, challenges such as the initial sensitivity of troponins and assay variability necessitate careful management to optimize diagnostic effectiveness.^17,18

Advancements in biosensor technologies have revolutionized the early detection of cardiovascular biomarkers, crucial for managing conditions such as AMI, where rapid diagnosis is vital due to the subtle and life-threatening nature of the symptoms. Electrochemical biosensors, known for their cost-effectiveness and high sensitivity, have been enhanced by nanomaterials and microfluidic integration, improving precision in cardiac marker detection.^20,21 Similarly, optical and microfluidic biosensors have increased diagnostic accuracy and achieved low detection limits through rapid, multiplex biomarker analysis.^22–24 Recent innovations in POC testing include surface-enhanced Raman scattering nanotags,²⁵ dedicated spectroscopy machines,²⁶ and fiber optic Nanogold-Linked Immunosorbent Assay,²⁷ along with novel strip that integrate fluorescence lateral flow assays for enhanced multi-target detection.^28,29 Additionally, ultra-rapid multitracking immunosensors have significantly reduced detection times for cardiac biomarkers, enhancing POC capabilities.^30,31 Despite these advancements, challenges such as standardization and ethical considerations continue to limit broader adoption.

To address high costs, limited portability, and scalability issues that hinder widespread clinical adoption, we developed THE ONE InstantCare platform, which integrates SpectroChip technology, also known Spectral Processing Unit (SPU), with LFIA technology (SPU × LFIA). This combination significantly enhances the quantitative analysis of cardiac biomarkers, providing a portable, cost-effective solution with rapid, accurate results, and improved limits of detection (LOD). In this study, THE ONE InstantCare platform was used to simultaneously quantify multiple cardiovascular biomarkers (cTnI, CK-MB, NT-proBNP, D-dimer, and CRP) using LFIA rapid test strips and was validated against hospital standards, demonstrating its potential as a reliable clinical tool.

2. Materials and methods

2.1. Sample preparation

In this study, we developed a standard curve and conducted clinical validation using two distinct sample preparation methods. For standard curve modeling, we spiked Fetal Bovine Serum (FBS; Gibco, Mexico) with a known concentration control solution (Multichem® S Plus; Technopath Manufacturing Ltd, Ireland). For clinical validation, we analyzed serum samples from 47 patients at Kaohsiung Municipal Siaogang Hospital (KMSH, Taiwan), targeting five cardiac biomarkers: cTnI (18 samples), CK-MB (4), NT-proBNP (9), D-dimer (8), and CRP (8). The broad concentration ranges provided a robust dataset to evaluate the performance of THE ONE InstantCare platform.

2.2. AuNPs conjugated LFIA test strip

In this study, we utilized colloidal gold nanoparticles (AuNPs) conjugated rapid test cassettes from Eternal Materials Co., Ltd Taiwan and Alltest Biotech Co., Ltd Hangzhou, China, designed to detect cardiac biomarkers. These test kits operate on AuNPs-based immunochromatographic assay, which uses anti-cardiovascular biomarker antibodies. During the test, patient sample antigens bind to these antibodies, and capillary action moves the mixture across a membrane, resulting in a visible red band where the intensity indicates biomarker concentration, as shown in Fig. 4. This enables both qualitative and semi-quantitative analyses, facilitating swift medical decisions. However, the interpretation of color changes can be subjective and lead to potential misinterpretations. To enhance accuracy, we employed THE ONE InstantCare platform, renowned for its ability to detect very low biomarker concentrations and improve result interpretation.

2.3. THE ONE InstantCare device: portable micro-spectrometer instrument

As shown in Fig. 1(a), THE ONE InstantCare technology is a compact (15 × 8 × 10 cm³) and precise micro-spectrometer instrument. It features advanced SpectroChip technology and a CMOS image sensor housed within an SPU module. Light is channelled through an optical fiber into the SPU where it is dispersed by an aberration-corrected concave micro-grating (AC-CMG) to optimize input and focus,^32,33 see Fig. 1(b). The spectrum is captured by the image sensor, processed with a 12-bit A/D converter to minimize stray light and noise levels (1.1% total noise), and output via USB for data analysis. Capable of operating across a 300–1000 nm spectral range with 5 nm resolution, it detects at sub-ppb sensitivity. Real-time signal processing within the SPU includes several key components: an adaptive Savitzky–Golay (SG) filter, which removes high-frequency noise while preserving peak integrity; dynamic baseline correction, which compensates for environmental drift; and automated gain control (AGC), which continuously adjusts detector gain relative to signal intensity to maintain an optimal signal-to-noise ratio (Fig. 3). The complete specifications of THE ONE InstantCare are detailed in Table S1. Designed for POC testing, this technology integrates seamlessly with clinical diagnostics, delivering rapid results within 15 minutes for various cardiac markers and enhances performance across medical, optical, environmental, and industrial settings with very low cost relative to other POCT devices.

Fig. 1 Micro-Spectrometer Integrated LFIA system representation. (a) THE ONE Instantcare device components, (b) Micro-Spectrometer Optical Pathway, and (C) relative absorbance (ΔA) spectral analysis of Aunps-based LFIA.

2.4. Spectral analysis

THE ONE InstantCare platform uses two white LEDs as a broad-spectrum light source, optimizing the measurement of light reflected from the cassette's reaction lines. The spectral data along the reaction line color intensity of known analyte concentration—typically the absorbance peaks—are used to construct the standard calibration curve. It is widely accepted that the peak absorbance of colloidal AuNP-based LFIA lies between 520 nm and 540 nm, with 530 nm commonly selected as the central wavelength.³⁴ To enhance accuracy and minimize background noise, the platform employs a specific analytical method, detailed in eqn (1), which calculates the relative absorbance (ΔA) by subtracting the absorbance at 680 nm from the peak absorbance within 500–600 nm, as shown in Fig. 1(c). As illustrated in Fig. 3, this method allows for the precise quantification of biomarker concentrations of unknown samples using a standard calibration curve, directly measuring the spectral intensity of colloidal AuNPs attached to antibodies in the sample.

ΔA = ΔA_{max(500–600 nm)} − A_{680 nm} (1)

2.5. Experimental procedure

In biomedical research, standard calibration curves are essential for quantitative analysis, enabling accurate determination of unknown sample concentrations by comparing experimental measurements, such as absorbance intensities, against known reference values. In this study, we prepared serial dilutions of standard FBS solutions and tested them multiple times using the proposed novel platform with rapid AuNP-based LFIA cassettes, obtaining stable average absorbance spectra at the test line with a very low coefficient of variation.³⁵ Relative absorbance (ΔA) values were calculated from the average spectral data at each concentration level, and calibration curves were constructed for five cardiac biomarkers, each measured in triplicate to ensure analytical reliability. These standard curves were then used by the ONE InstantCare system to accurately quantify antibody concentrations in unknown samples.

The general procedural workflow of the proposed platform is illustrated in Fig. 2, outlining the operational steps of THE ONE InstantCare quantitative cardiac biomarker analysis. The process begins with the application of the sample onto the LFIA rapid test kit, followed by the addition of a buffer reagent to initiate the reaction. As the sample migrates along the strip, visible lines form in the test region, indicating the presence or absence of the target analyte. The strip is then inserted into THE ONE InstantCare device, which uses advanced spectrometry to capture high-resolution images and analyze light absorbance at the reaction line. The device's signal processing unit (SPU) interprets the spectral data and applies a pre-established, robust standard curve to accurately quantify the analyte concentration, delivering both qualitative and quantitative results (see Fig. 3). This method combines the simplicity of LFIA with the precision of spectrometry, enabling rapid and accurate point-of-care testing.

Fig. 2 Experimental procedure: illustration of THE ONE instantcare platform for key cardiovascular biomarkers quantification.

Fig. 3 Workflow of SPU-based spectral data processing for quantitative biomarker analysis using THE ONE Instantcare system.

For validation, we analyzed serum samples obtained from KMSH Hospital. These samples were first quantified using the hospital's gold-standard analyzer, the Abbott Alinity ci-series clinical chemistry and immunology system, to establish reference concentrations. Subsequently, the same samples were tested with THE ONE InstantCare system, which captured absorbance spectra at the reaction line. Relative absorbance (ΔA) values were computed from the average spectral data of each unknown sample. Using these ΔA values and the previously established calibration curves, the concentrations of the unknown biomarkers were determined. To evaluate the performance of THE ONE InstantCare system relative to the hospital reference, we employed Bland–Altman analysis and Pearson's correlation. The statistical comparisons confirmed that the ONE InstantCare device reliably quantifies cardiovascular biomarkers, demonstrating its effectiveness as a point-of-care (POC) diagnostic tool.

2.6. Statistical analysis

2.6.1. Limit of detection (LOD). In analytical chemistry and diagnostics, the limit of detection (LOD) and limit of quantification (LOQ) are essential measures of an analytical method's sensitivity and reliability. LOD refers to the lowest concentration of a substance that can be reliably detected, whereas LOQ represents the lowest concentration that can be accurately quantified. Different methods for calculating these limits are tailored to specific applications and considerations, underscoring the importance of choosing appropriate approaches to determine these critical thresholds.

■ Method-I: calibration curve method

The IUPAC recommends calculating the LOD by using the standard deviation (SD) of the response and the slope of the calibration curve near the LOD level, as detailed in eqn (2). Assuming a normal distribution, the LOD is defined as the concentration at which an analyte can be detected from the blank with 99% certainty.

where: σ is the standard deviation of the response, and S is the slope of the calibration curve. Factor 3.3, ensures a conservative estimate for the LOD at a 99% confidence level, covering random and systematic errors.

■ Method-II: blank measurement method

This method calculates the LOD using the mean and standard deviation of the blank. It focuses on distinguishing the smallest measurable signal from the background noise of the assay. This is particularly useful in assays where the blank may exhibit a nonzero signal due to background noise or interference. Assuming a normal distribution, any signal exceeding this calculated value can be confidently distinguished from the blank with 99.7% confidence.

LOD = blank_mean + 3.3 × blank_SD (3)

where, blank_mean and blank_SD are average absorbance and SD of the blank samples, respectively.

2.6.2. Limit of quantification (LOQ). Essentially, the LOQ is a critical analytical parameter that indicates the concentration at which results are reliably quantitative for analysis and decision-making in various fields. Similarly, we consider two potential methods for calculating LOQ.

■ Method-I: calibration curve method

As presented in eqn (4), the formula of LOQ uses the slope of the calibration curve and the standard deviation of the response with a factor of 10 ensuring a conservative estimate at a 95% confidence interval, and expressed as follows:

where σ is the SD of the response and S is the slope of the calibration curve.

■ Method-II: blank measurement method

This method depends on blank measurements and typically incorporates a factor to ensure adequate signal strength and acceptable variability, as expressed in eqn (5). A common formula used to calculate the LOQ, particularly in methods like spectroscopy where blank measurements influence the analysis, is expressed as:

LOQ = blank_mean + 10 × blank_SD (5)

2.6.3. Pearson's correlation analysis. Pearson's correlation coefficient, known as Pearson's r, is a statistical tool that measures the linear relationship between two variables, ranging from −1 (i.e., perfect negative linear correlation) to +1 (i.e., perfect positive linear correlation) and a value of 0 indicates no correlation. Common thresholds, r < 0.25 for no correlation, 0.25 < r < 0.5 for weak, 0.5 < r < 0.75 for moderate, and r > 0.75 for strong, are used to categorize the strength of the correlation, though these are not absolute and can vary by field of study.³⁶ This study utilized Pearson's r to evaluate the correlation between the ONE InstantCare platform and the Abbott Alinity ci analyzer. It is a crucial approach to determine the strength and direction of the relationship between these two diagnostic tools.

2.6.4. Bland–Altman analysis. Bland–Altman analysis, known as a method of difference, is a statistical approach commonly used in clinical and biomedical research to evaluate the agreement between two diagnostic tools that intend to measure the same parameter. It calculates the differences and averages of paired measurements from the two methods and then plots these differences against the averages to analyze the agreement visually. The plot shows the mean difference and limits of agreement, defined as the mean difference plus and minus 1.96 times the standard deviation of the differences.³⁷ In this study, the Bland–Altman analysis evaluated the agreement between data from THE ONE InstantCare and the Abbott Alinity ci analyzer, using a 95% confidence interval to establish upper and lower limits. This method identifies potential bias and precision discrepancies, determining if two methods can be reliably interchanged.

3. Results

3.1. Standard curve modelling

A control solution of known concentration was added to FBS and serially diluted to prepare the samples. Using THE ONE InstantCare platform, we accurately detected the absorption spectra of the test lines (T-lines) on the test cassettes. As shown in Fig. 4, the average absorbance spectra from multiple cardiovascular LFIA rapid test cassettes measured independently three times to ensure consistency and robustness. A pronounced absorption peak at 530 nm, typical of colloidal AuNPs, is evident in these spectral data. The intensity of these spectral peaks correlates directly with the darkness of the detection test lines on the cassettes, indicating higher concentrations of bound antibodies correspond to stronger spectral peaks.

Fig. 4 Average absorbance spectral data of key cardiovascular biomarkers at different concentration values: (a) cTnI, (b) CK-MB, (c) NT-ProBNP, (d) D-dimer, and (e) CRP.

The relative absorbance (ΔA) was determined for each spectral measurements using eqn (1), and linear standard calibration curves were constructed by plotting ΔA against the biomarker concentration. Fig. 5 illustrates these modelled standard curves of clinically relevant cardiac biomarkers.

Fig. 5 Linear standard curve model of cardiac biomarker: (a) cTnI, (b) CK-MB, (c) NT-proBNP, (d) D-dimer, (e) CRP.

3.2. LOD and LOQ

3.2.1. Method-I: calibration curve method. As discussed in section 2.6, Method-I estimates LOD and LOQ based on the calibration curve's slope and the SD of response residuals, i.e., the difference between actual and predicted responses (y_actual, y_pred), as shown in eqn (6) and visualized in Fig. S2 of the SI. The high linearity of the calibration curves (R² > 0.98 for all biomarkers) demonstrates the system's strong quantitative predictive performance. Key performance metrics, including slope, SD, LOD, and LOQ, are summarized in Table 1.

Residual = y_actual − (mx + b) (6)

Table 1 Performance summary of five key cardiovascular biomarkers analyzed using THE ONE InstantCare platform, calibration curve linearity and LOD/LOQ results estimated by two different approaches

Analytical Parameters	cTnI	CK-MB	NT-proBNP	D-dimer	CRP
R ^2	0.9824	0.9924	0.9875	0.9872	0.9941
Slope	0.0574	0.0117	0.0483	0.1993	0.0638
Meanblank	0.0003	0.0009	0.0020	0.0008	0.0006
SDresponse	0.0060	0.0106	0.0265	0.0199	0.0172
SDblank	0.0005	0.0017	0.0040	0.0086	0.0012
LODmethod−I [ng mL^−1]	0.3429	2.9723	1.8070	0.3301	0.8877
LODmethod−II [ng mL^−1]	0.0124	0.0342	0.1422	0.0213 × 10^−3	0.0304 × 10^−3
LOQmethod−I [ng mL^−1]	1.0391	9.0069	5.4758	1.0002 × 10^−3	2.6901 × 10^−3
LOQmethod−II [ng mL^−1]	0.0256	0.0828	0.2525	0.0561 × 10^−3	0.0724 × 10^−3
Linear range, LLOQ – ULOQ [ng mL^−1]	0.525–2.1	1.875–30.0	0.234–15.026	(0.0–2.48) × 10^−3	(0.63–20.31) × 10^−3

---

3.2.2. Method-II: blank measurement method. Method-II estimates LOD and LOQ from the statistical properties of blank spectral data (i.e., sample with zero analyte concentration), specifically, the mean (blank_mean) and SD (blank_SD) of baseline signals. To improve precision, baseline correction was applied to the raw blank spectra (Fig. S3) using a rolling minimum subtraction technique (Fig. S4), effectively reducing instrumental drift and spectral noise. This preprocessing step ensured a more stable baseline, increasing the reliability of sensitivity assessments.

The analytical sensitivity outcomes from both methods are summarized in Table 1. All five cardiac biomarkers demonstrated low LOD and LOQ values, affirming the platform's capability to detect low-concentration analytes. Notably, biomarkers exhibited the lowest LOD and LOQ under Method-II, underscoring the high precision achieved through blank spectrum-based estimation. These findings support the effectiveness of the SpectroChip-integrated LFIA system for high-resolution, and quantitative detection of cardiovascular biomarkers.

3.3. Clinical validation

To assess the clinical reliability of our platform, we compared spectral measurement results from human serum samples with those obtained using the Abbott Alinity ci clinical analyzer. As presented in Table 2, the comparison showed strong agreement across all five cardiac biomarkers. Notably, CK-MB, NT-proBNP, D-dimer, and CRP exhibited near-perfect alignment, with R² values above 0.97 and Pearson's r exceeding 0.98. Even for cTnI, the correlations remained robust (R² = 0.9187), demonstrating consistent performance across both high- and low-abundance targets.

Table 2 Correlation metrics (R² and Pearson's r) comparing THE ONE Instantcare platform and the Abbott Alinity Ci Analyzer for five key cardiovascular biomarkers based on serum sample analysis

Parameters	cTnI	CK-MB	NT-proBNP	D-dimer	CRP
R ^2	0.9187	0.9999	0.9910	0.9748	0.9810
Pearson's r	0.9585	1.0000	0.9955	0.9873	0.9905

---

Further validation is shown in Fig. 6, which presents Bland–Altman plots confirming the absence of significant bias between the two methods across all biomarkers. Complementing this, Fig. 7 illustrates the Pearson's correlation analysis, visually reinforcing the high level of agreement. Together, these results support the accuracy and clinical applicability of THE ONE InstantCare platform for quantifying cardiac biomarkers in real-world patient samples.

Fig. 6 Bland–Altman analysis of THE ONE InstantCare Platform with Abbott Alinity ci Analyzers for key cardiac biomarkers quantitative analysis: (a) cTnI, (b) CK-MB, (c) NT-proBNP, (d) D-dimer, and (e) CRP.

Fig. 7 Pearson's correlation analysis of THE ONE InstantCare platform with abbott alinity ci analyzers demonstrating linear associations and statistical significance: (a) cTnI, (b) CK-MB, (c) NT-proBNP, (d) D-dimer, and (e) CRP.

3.3.1. Troponin I (cTnI). cTnI, a key biomarker for diagnosing myocardial infarction, was analyzed using 18 serum samples with varying concentrations. Fig. 6(a) compares two assay results, the spectral measurements had an average bias of 0.006 ng mL⁻¹ and an SD of 0.847 ng mL⁻¹. Bland–Altman analysis showed that 17 of these samples fell within the 95% confidence interval, with just one exceeding the lower limit (+1.96 SD). As illustrated in Fig. 7(a), with Pearson's r at 0.9585 and an R² at 0.9187, these results confirm the comparability of cTnI measurements across both methods.

3.3.2. Creatine kinase-myocardial band (CK-MB). CK-MB, an enzyme crucial for diagnosing myocardial infarction, was analyzed using 4 serum samples. The comparative results, shown in Fig. 6(b), reveal a mean spectral bias of −0.175 ng mL⁻¹ and an SD of 0.7457 ng mL⁻¹. According to Bland–Altman analysis, all samples were within the 95% confidence interval. As shown in Fig. 7(b), the extremely high Pearson's r value of 1.0000 and an R² of 0.9999 indicate strong agreement between the two measurement methods, confirming the reliability of the CK-MB measurements.

3.3.3. N-Terminal pro-B-type natriuretic peptide (NT-proBNP). NT-proBNP, an essential biomarker for diagnosing and managing heart failure, was analyzed using 9 serum samples. Fig. 6(c) displays the comparative results from the two methods, showing a mean concentration bias of 0.9265 ng mL⁻¹ and an SD of 0.8993 ng mL⁻¹. According to Bland-Altman analysis, all samples were within the 95% confidence interval. As depicted in Fig. 7(c), Pearson's r of 0.9955 and R² of 0.9910, confirm the strong agreement between the two methods, which validates the reliability of the proposed platform for NT-proBNP measurements.

3.3.4. D-Dimer. D-Dimer, a protein fragment essential for diagnosing thrombosis-related disorders like deep vein thrombosis and pulmonary embolism, was assessed in 8 serum samples. Fig. 6(d) summarizes the results from two measurement methods, showing a mean concentration bias of −0.017 mg L⁻¹ and an SD of 0.3906 mg L⁻¹. Bland–Altman analysis revealed that seven samples fell within the 95% confidence interval, with only one slightly outside the lower limit (−1.96 SD). With a Pearson's r of 0.9873 and an R² of 0.9748, the findings confirm that the D-dimer measurements are consistent across both methods as shown in Fig. 7(d).

3.3.5. C-reactive protein (CRP). CRP, an acute-phase protein indicative of inflammation and a crucial biomarker for CVD, was tested in 8 serum samples. Comparative results presented in Fig. 6(e) show a mean concentration bias for CRP of −1.035 mg L⁻¹ and an SD of 1.854 mg L⁻¹. All samples remained within the 95% confidence interval according to Bland–Altman analysis. With a Pearson's r of 0.9905 and an R² of 0.9810, the results demonstrate substantial agreement between the two measurement methods, confirming their comparability for CRP assessments, see Fig. 7(e).

3.4. Linear range

The linear range, also known as the analytical measuring range (AMR), is the span of concentrations over which the assay response (ΔA at 530 nm) remains directly proportional to analyte concentration. Within this interval, the calibration curve maintains acceptable linearity (R² ≥ 0.98) and precision. Outside this range, very low concentrations are dominated by noise, while higher concentrations may show signal saturation or nonlinearity. In this study, the lower-LOQ (LLOQ) was defined as the lowest concentration meeting two criteria: (i) coefficient of variation (CV%) ≤ 20% across replicates (n = 3), and (ii) residual error within ±20% of the nominal value. The upper-LOQ (ULOQ) was taken as the highest concentration maintaining acceptable linearity and precision. Following regulatory guidance from the U.S. Food and Drug Administration (FDA), the European Medicines Agency (EMA), and the Clinical and Laboratory Standards Institute (CLSI), we applied thresholds of CV% ≤ 20% at the LLOQ and ≤15% at other concentrations.

A practical approach was used to confirm the linear range: the blank sample was employed to estimate background and the LOD, while low-end quantification was empirically validated using replicate samples near the estimated LOD. The final linear ranges for all biomarkers were reported as [LLOQ–ULOQ] (Table 1). Because absorbance values near zero are highly susceptible to small fluctuations, the CV% at the blank was expectedly high. For this reason, blanks were generally excluded from precision analysis and used only for background estimation. Notably, however, for D-dimer, the blank sample (0.0 ng mL⁻¹) consistently generated a reproducible and distinguishable signal relative to baseline. This unusual precision and accuracy at the zero-concentration point allowed us to confidently assign the LLOQ for D-dimer at 0.0 ng mL⁻¹—a feature not observed for the other biomarkers, and one that highlights the platform's potential for highly sensitive detection in clinically relevant settings.

4. Discussion

In this study, we employed colloidal AuNPs-based LFIA and performed a comprehensive statistical evaluation of spectral data obtained at the test line of the assay. As illustrated in Fig. 8, this analysis is critical for optimizing assay design, particularly in selecting wavelengths that minimize signal variability and enhance the accuracy of biomarker quantification. The figure presents the standard deviation (SD) of averaged absorbance spectra across different biomarker concentrations, revealing notably higher variability within the 500–600 nm range—a key spectral window for analyte detection. This elevated SD in the target region indicates active signal fluctuation corresponding to biomarker presence, validating its diagnostic relevance. Conversely, the lower SD values outside this range suggest limited responsiveness, reinforcing the specificity of the 500–600 nm region.³⁵

Fig. 8 Analysis of spectral variability in cardiac biomarker concentrations: standard deviation (SD) across biomarkers concentration.

For cTnI, the spectra exhibited relatively more fluctuation compared to other biomarkers, including secondary peaks around 450 nm and 650–700 nm. These features can be attributed to nonspecific background scattering from the nitrocellulose membrane and optical effects such as Mie scattering and higher-order plasmon resonance modes of colloidal AuNP conjugates, particularly at low analyte concentrations.^38–40 Similar spectral artifacts have been reported in the literature, where AuNP aggregation or resonance coupling can generate secondary peaks beyond the main localized surface plasmon resonance (LSPR) band.^41,42 Crucially, these peaks are not directly related to antigen–antibody binding and therefore do not affect the quantification of cTnI. Quantitative calibration was consistently performed at 530 nm, where the LSPR of AuNPs produces the strongest and most reliable optical response.^39,40 The high linearity (R² > 0.98) achieved in this range confirms that background-induced fluctuations outside the main LSPR band do not bias quantification, supporting the robustness and reliability of the proposed spectrometer-integrated LFIA platform.

As shown in Fig. 5, the SD bars indicate that the variability in average absorbance measurements increases with concentration. This trend, also visible across all biomarkers in Fig. 4, underscores the concentration-dependent precision of the assays. Furthermore, the implementation of one-factor ANOVA on the average spectral data of all biomarkers resulted in an extremely small P-value, approximating zero, which is not practically achievable in an absolute sense; it is commonly understood and accepted in the scientific community as indicating a result where the evidence against the null hypothesis is robust (see Tables S3–S7). This demonstrates that the spectral analysis is highly sensitive to changes in concentration, confirming that our protocol can effectively differentiate between various concentrations of the sample. Lower concentrations display smaller variability, indicating greater precision with the LFIA method. Conversely, higher concentrations exhibit increased SDs, highlighting possible performance limitations at these levels.

The spectral data illustrated in Fig. 4 demonstrate the dependency of absorbance on varying analyte concentrations. To ensure a comprehensive evaluation of the analytical sensitivity of our platform, we employed two complementary approaches for LOD and LOQ estimation. Method I, the calibration curve-based approach, relies on the slope of the regression line and the standard deviation of the response. This method is widely adopted in bioanalytical validation because it reflects real-world assay behavior across measured low-concentration samples. Method II, the blank-based approach, leverages the spectral variability of blank samples, measured by the proposed micro-spectrometer system, to highlight the instrument's theoretical detection capacity under minimal noise conditions.

Table 1 summarizes the analytical performance of THE ONE InstantCare platform using both approaches. Notably, the two methods yield distinct values across the five cardiac biomarkers, reflecting differences in sensitivity and baseline noise characteristics. Method I (eqn (2) and (4)), which assumes a stable linear response and consistent variability near the low concentration range, generally provides more conservative estimates and is therefore the more clinically realistic measure of performance. By contrast, Method II (eqn (3) and (5)) significantly lowers detection thresholds by exploiting minimal SD at the blank, illustrating the platform's maximal theoretical sensitivity. However, such estimates may be optimistically low under ideal conditions and should not replace empirical validation. A blank-based approach is appropriate to estimate the LOD, but the LOQ must be demonstrated with measured low-concentration samples that satisfy precision–accuracy criteria. In this study, as presented in the linear range (see Table 1), LOQ was empirically determined by preparing multiple (n = 3) independent replicates at low non-zero concentrations and accepting the concentration as the lower LOQ (LLOQ) if the CV was ≤20%.

Taken together, the use of both methods provides a balanced framework: Method I reflects practical and clinically relevant quantification performance, while Method II underscores the instrument's fundamental detection capacity. This dual evaluation highlights both the robust applicability of our platform in clinical diagnostics and its technical sensitivity ceiling, which may be further exploited in controlled laboratory or research environments.

As shown in Table 3, current clinical diagnostic thresholds for cardiac biomarkers—such as cTnI (≥0.04–0.5 ng mL⁻¹ for MI), NT-proBNP (0.45–1.8 ng mL⁻¹ for HF), CRP (>3 mg L⁻¹ for high cardiovascular risk and >10 mg L⁻¹ for inflammation), CK-MB (≥5–10 ng mL⁻¹ for MI), and D-dimer (≥0.5 μg mL⁻¹ for VTE)—highlight the need for highly sensitive assays capable of detecting early or subclinical disease. Traditional calibration-based approaches (Method-I), often lack sufficient sensitivity near these clinical thresholds. For example, conventional assays may miss early-stage MI or HF when cTnI or NT-proBNP levels are just below their diagnostic cutoffs. In contrast, Method-II leverages the spectral variability of blank samples, uniquely measured using our micro-spectrometer system, to achieve ultra-low LOD values—such as 0.0124 ng mL⁻¹ for cTnI and 0.1422 ng mL⁻¹ for NT-proBNP (see Table 1)—which are 3–100× more sensitive than required clinical thresholds. This enhanced sensitivity enables early biomarker detection, reduces false negatives, and facilitates accurate quantification even at low concentrations, as confirmed by corresponding lower limit of quantification (LLOQ) values within the linear range (see Table 1). Method-II's robustness is further supported by its minimal SD_blank, ensuring reproducibility. Collectively, this method bridges the gap between analytical performance and clinical utility, offering transformative potential for early diagnosis, risk stratification, and personalized cardiovascular care.

Table 3 Clinical reference ranges, diagnostic thresholds, and interpretive insights for key cardiovascular biomarkers (cTnI, CK-MB, NT-proBNP, D-dimer and CRP)

Biomarker	Normal Range	At-Risk/Diagnostic Threshold	Clinical Use	Remark
URL: upper reference limit; HF: heart failure; MI: myocardial infarction; AMI: acute myocardial infarction; AF: atrial fibrillation; FEU: fibrinogen equivalent units; DDU: D-dimer Units; VTE: venous thromboembolism; DVT: deep vein thrombosis; PE: pulmonary embolism; CVD: cardiovascular disease; and DIC: disseminated intravascular coagulation.
Troponin I (cTnI)	0.011–0.0429 ng mL^−1	≥0.04–0.5 ng mL^−1 (99^th percentile URL)	MI detection	Vary by assay, sex and population.^46,47
CK-MB	≤5 ng mL^−1 or 3–5% of total CK	>5–10 ng mL^−1 or >5% of total CK (99^th percentile URL)	MI detection	Vary by assay and laboratories.^48,49
NT-proBNP	<0.125 ng mL^−1 (<75 years)	>0.450 ng mL^−1 (<50 years)	HF & Cardiac stress	Vary by age, sex, renal function, obesity, AF, medications and other clinical contexts^50–54
	<0.450 ng mL^−1 (≥ 75 years)	>0.900 ng mL^−1 (50–75 years)
		>1.8 ng mL^−1 (>75 years)
D-dimer	<0.5 μg mL^−1 (FEU) or <500 ng mL^−1 (DDU) in non-pregnant adults	>0.5 μg mL^−1 FEU (≤50 years)	VTE (DVT/PE), Coagulation	An elevated level of D-dimer is non-specific & caused by numerous conditions (advanced age, pregnancy, inflammation, infection, recent surgery, trauma, AMI, liver disease, cancer, stroke, DIC) other than VTE. Vary by assay, has low specificity, and not used for high pre-test probability^55–62
		Age × 0.01 μg mL^−1 (>50 years)
CRP	<1 mg L^−1 (low risk)	1–3 mg L^−1 (average risk of CVD); > 3 mg L^−1 (high risk); & >10 mg L^−1 (inflammation or infection)	Inflammation, CVD risk	Elevated CRP indicates systemic inflammation and is associated with an increased risk of atherosclerosis, myocardial infarction, and stroke. CRP is a non-specific marker of inflammation and can be elevated due to various conditions beyond cardiovascular risk, such as infections, trauma, chronic inflammatory diseases (e.g., rheumatoid arthritis), obesity, diabetes, and even smoking.^63–68

---

Fig. 4 highlights the practical challenges of visually discerning low-concentration reaction lines on LFIA strips, particularly when the T-line intensity approaches that of the C-line. This limitation is inherent to conventional qualitative LFIA methods, where interpretation often depends on the human eye or simple optical readers, and low analyte concentrations may be overlooked or misclassified.

Table 4 provides a comparative analysis that situates our SpectroChip-based THE ONE InstantCare platform alongside widely used POCT devices. Importantly, the table demonstrates that our platform achieves lower limits LoD than traditional LFIA approaches, effectively overcoming the visual detection barrier highlighted in Fig. 4. This very high analytical sensitivity is a defining advantage of the system, ensuring that even subtle biomarker elevations can be reliably detected.

Table 4 Comparative analytical performance and operational characteristics of commercially available point-of-care testing (POCT) platforms for cardiac biomarkers

Device	Size/weight	Target analyte	Sample size (type)	Turnaround time	LOD	Linear range (AMR)	Cost & usability	Reference
THE ONE InstantCare	15 × 8 × 10 cm, 490 g	cTnI, CK-MB, NT-proBNP, D-dimer, CRP (depending on strip)	10–20 μL (Whole blood/serum via strip) – Fingersticks	5–10 min	cTnI: 0.0124 ng mL^−1; CK-MB: 0.0342 ng mL^−1; NT-proBNP: 0.1422 ng mL^−1; D-dimer: 0.0213 ng L^−1; CRP: 0.0304 ng L^−1	cTnI: 0.525–2.1 ng mL^−1; CK-MB: 1.875–30.0 ng mL^−1; NT-proBNP: 0.234–15.026 ng mL^−1; D-dimer: 0.0–2.48 ng L^−1; CRP: 0.63–20.31 ng L^−1	Low-cost; portable reader; minimal training; easy use; affordable & easily accessible.	This study.
Conventional LFIA	Handheld strip only, lightweight	All AuNP-LFIA (depending on the target)	10–20 μL (whole blood/Serum via strip)	10–15 min	cTnI: 0.50 ng mL^−1; CK-MB: 5 ng mL^−1; NT-proBNP: 0.50 ng mL^−1; D-dimer: 0.05 ng L^−1; CRP: 1 ng L^−1	None (qualitative approach of detection)	Very lightweight	Eternal Materials and Alltest Biotech User Guidelines.
Roche cobas® h 232	24.4 × 10.5 × 5.1 cm, 526 g	cTnI, CK-MB, NT-proBNP, D-dimer, Myoglobin (depending on cartridge)	50–150 μL (Whole blood/serum)	8–12 min	Not specified	cTnI: 0.04–2 ng mL^−1; CK-MB: 1–40 ng mL^−1; NT-proBNP: 0.06–9 ng mL^−1; D-dimer: 0.1–4.0 μg mL^−1; Myoglobin: 30–700 ng L^−1	Moderate cost; easy use, CLIA-waived; single-test strips	Roche Diagnostics IFU (Instructions for Use)
Abbott i-STAT®	23.48 × 7.68 × 7.24 cm, 650 g	cTnI, CK-MB, NT-proBNP, D-dimer (depending on cartridge)	20–100 μL (whole blood/serum)	5–10 min	cTnI: 0.006 ng mL^−1; CK-MB: 0.006 ng mL^−1; NT-proBNP: 0.0047 ng mL^−1; D-dimer: 0.068 ng L^−1 (FEU)	cTnI: 0.012–50.0 ng mL^−1; CK-MB: 0.1–15.0 ng mL^−1; NT-proBNP: 0.005–4.5 ng mL^−1; D-dimer: 0.09–8.50 ng L^−1 (FEU)	High device cost; higher cartridge cost; requires a trained operator	Abbott i-STAT Technical Guide
Samsung LABGEO IB10	17.5 × 33×17.7 cm, 2.4 kg	cTnI, CK-MB, D-dimer, NT-proBNP, Myoglobin	500 μL (whole blood/Plasma)	20 min	Not specified	cTnI: 0.05–30 ng mL^−1; CK-MB: 2–60 ng mL^−1; NT-proBNP: 0.03–5.0 ng mL^−1; D-dimer: 100–4000 ng mL^−1 (FEU); Myoglobin: 30–500 ng mL^−1	Larger benchtop; higher cost; cartridges moderate; requires lab setting	Samsung LABGEO IB10 IFU

---

It should be noted that the AMRs reported for our platform are defined by the lowest concentrations measured with acceptable precision (CV <20%). These values do not necessarily represent the absolute LLOQ, but rather the concentrations that can be consistently measured and reported with confidence. In practice, this means our system can detect analytes closer to baseline than many existing POCT devices, even if not formally reported as LLOQ.

By integrating high sensitivity with portability, rapid turnaround (5–10 minutes), and minimal sample requirements (10–20 μL whole blood), the InstantCare system demonstrates performance at least comparable to commercial POCT devices such as the Roche cobas® h 232, Abbott i-STAT®, and Samsung LABGEO IB10—while offering greater sensitivity and practical advantages. This positions our platform as a clinically realistic and more powerful alternative for point-of-care cardiac biomarker testing, particularly in urgent and decentralized care settings where early detection of low biomarker concentrations is critical.

The comparative performance metrics presented in Table 2 demonstrate strong agreement between THE ONE InstantCare system and the Abbott Alinity ci-series clinical analyzer, supporting the diagnostic potential of our proposed platform. Across all five key cardiac biomarkers tested, THE ONE InstantCare achieved exceptionally high Pearson's correlation coefficients (r > 0.95) and coefficients of determination (R² > 0.91), indicating near-perfect linear relationships between the two measurement systems, as shown in Fig. 7. These strong correlations highlight the analytical precision and consistency of the micro-spectrometer-based platform relative to gold-standard laboratory instruments. Bland–Altman analysis further confirms this agreement, with measurement differences consistently falling within the acceptable ±1.96 standard deviation limits and no evidence of systematic bias (see Fig. 6). Together, these findings suggest that THE ONE InstantCare system delivers accurate, reproducible, and clinically reliable results across a wide range of biomarker concentrations. This level of performance positions the system as a viable solution for decentralized or point-of-care settings, where rapid and dependable diagnostics are essential.

THE ONE InstantCare system enhances the precision of LFIA rapid test strips for quantitative biomarker analysis, vital for diagnosing and treating cardiovascular diseases (CVDs) promptly. Ideal for urgent care settings such as ambulances and emergency rooms, it enables rapid and precise biomarker analysis without specialized professionals, facilitating immediate patient care. This platform quickly and accurately measures key cardiovascular biomarkers, cTnI, CK-MB, NT-proBNP, D-dimer, and CRP, matching the performance of gold-standard medical equipment. Its compact design, ease of use, and cost-effectiveness make it suitable for use in smaller medical facilities and remote locations lacking traditional lab facilities.^43–45 By enabling timely interventions, THE ONE InstantCare system can significantly improve patient outcomes and transform cardiovascular healthcare.

4.1. Analytical scope and limitations

For D-dimer, our platform successfully distinguishes blank samples (0 ng mL⁻¹) from true signals with good reproducibility. The first measurable non-zero concentration tested also met the precision and accuracy criteria, which allows it to be reported as the effective lower limit of quantification (LLOQ).

For other biomarkers, however, the empirically determined linear ranges sometimes begin above their clinical diagnostic cut-offs. This means that while THE ONE InstantCare demonstrates very high sensitivity (reflected in low limits of detection, LODs) and strong correlation with laboratory-based analyzers, its quantifiable range (LLOQ → ULOQ) does not always extend into the very lowest concentrations used in clinical decision-making.

This distinction highlights two complementary aspects of performance:

1. Detection sensitivity (LOD): the ability to detect signal above background. Here, our platform frequently outperforms existing LFIA readers.

2. Quantitative accuracy (LLOQ): the ability to reliably report concentrations with acceptable precision and bias. This defines the clinically reportable range.

We report both LOD/LOQ and linear ranges transparently (Table 1) and compare them with representative commercial POCT LFIA readers (Table 4). In cases where the LLOQ is above a clinical decision threshold, the implication is clear: the platform can detect early or subclinical signals, but additional calibration, replicate testing, or assay optimization may be needed to quantify these concentrations reliably. Further clinical validation is therefore recommended.

A specific limitation also arises for CK-MB. As shown in Fig. 6(b) and 7(b), validation was based on only four clinical samples. While the resulting R² value of 1 indicates perfect linear correlation, this finding must be interpreted cautiously. Small sample sizes inherently reduce statistical power and may mask biological or technical variability. The observed high correlation may reflect sample homogeneity or the controlled dataset, rather than real-world performance. Expanding validation with a larger, more diverse patient population is essential to confirm the generalizability of CK-MB results.

In general, the main study limitations include:

• Small clinical sample size for certain biomarkers.

• Limited replicate measurements at very low concentrations, which restricts precise mapping of the platform's LLOQ.

• Use of a linear calibration model, which may oversimplify assay behavior at low and high ends of the curve.

Future studies with expanded datasets, replicate testing near the LLOQ, and nonlinear calibration models could substantially improve the accuracy and robustness of the platform's validation.

5. Conclusions

In this study, THE ONE InstantCare platform—featuring SpectroChip technology integrated with colloidal AuNPs-based LFIA—demonstrated high accuracy and sensitivity in the detection and quantification of key cardiovascular biomarkers. Validation against the gold-standard Abbott Alinity ci-series clinical analyzer confirmed the system's reliability and clinical relevance. With its compact design (15 × 8 × 10 cm), rapid analysis time (5–10 minutes), and cost-effectiveness, the platform is particularly well-suited for emergency and POC settings, where timely decision-making is critical. Its advanced spectroscopic capability enables precise quantification at very low biomarker concentrations, achieving limits of detection 3–100× lower than conventional LFIA approaches. These improvements make the platform highly effective for early diagnosis and monitoring of cardiovascular disease (CVD), heart failure (HF), inflammation, venous thromboembolism (VTE), and other cardiac conditions. Overall, the ONE InstantCare system offers a scalable and adaptable diagnostic solution that could significantly enhance accessibility, accuracy, and speed in modern clinical practice.

Author contributions

C.-H. Ko: conceptualization, methodology, funding acquisition, project administration, resources, writing – review & editing, supervision, validation, funding acquisition; W.-Y. Kong: conceptualization, data curation, methodology, software, formal analysis, investigation, visualization, writing – original draft.

Conflicts of interest

There are no conflicts to declare.

Ethics approval

This study was approved by the Kaohsiung Municipal Siaogang Hospital (KMSH) Ethics Committee (IRB No. 25MMHIS184e) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to sample collection.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Supplementary information includes. 1. Raw data of the sample. spectrum tested by the machine. 2. The data tested by the hospital machine. 3. Sample photo. See DOI: https://doi.org/10.1039/d5an00846h.

Acknowledgements

This research was supported by internal resources from National Taiwan University of Science and Technology (NTUST, Taipei City, Taiwan), SpectroChip Inc. (Hsinchu City, Taiwan), and Kaohsiung Municipal Siaogang Hospital (KMSH, Kaohsiung City, Taiwan). No external funding was received.

References

1. O. Gaidai, Y. Cao and S. Loginov, Global Cardiovascular Diseases Death Rate Prediction, Curr. Probl. Cardiol., 2023, 48, 101622.
2. V. Lopes, J. P. Baptista, N. Moreira and L. Goncalves, Cardiovascular disease in critically ill COVID-19 patients, Eur. Hear. J Acute Cardiovasc. Care, 2022, 11 DOI:10.1093/ehjacc/zuac041.021.
3. A. Hessami, A. Shamshirian, K. Heydari, F. Pourali, R. Alizadeh-Navaei, M. Moosazadeh, S. Abrotan, L. Shojaie, S. Sedighi, D. Shamshirian and N. Rezaei, Cardiovascular diseases burden in COVID-19: Systematic review and meta-analysis, Am. J. Emerg. Med., 2021, 46, 382–391.
4. D. A. Watkins, Cardiovascular health and COVID-19: Time to reinvent our systems and rethink our research priorities, Heart, 2020, 106, 1870–1872.
5. R. T. Bhatia, S. Gati, M. Papadakis and S. Sharma, The Impact of COVID-19 on the Continuity of Cardiovascular Care, Eur. Heart J., 2021, 42, 215–217.
6. G. A. Roth, M. Vaduganathan and G. A. Mensah, Impact of the COVID-19 Pandemic on Cardiovascular Health in 2020: JACC State-of-the-Art Review, J. Am. Coll. Cardiol., 2022, 80, 631–640.
7. M. A. Faghy, J. Yates, A. P. Hills, S. Jayasinghe, C. daLuz Goulart, R. Arena, D. Laddu, R. Gururaj, S. K. Veluswamy, S. Dixit and R. E. M. Ashton, Cardiovascular disease prevention and management in the COVID-19 era and beyond: An international perspective, Prog. Cardiovasc. Dis., 2023, 76, 102–111.
8. U. Friess and M. Stark, Cardiac markers: A clear cause for point-of-care testing, Anal. Bioanal. Chem., 2009, 393, 1453–1462.
9. M. Shang, J. Guo and J. Guo, Point-of-care testing of infectious diseases: recent advances, Sens. Diagn., 2023, 2, 1123–1144.
10. D. A. Morrow and J. A. DeLemos, Benchmarks for the assessment of novel cardiovascular biomarkers, Circulation, 2007, 115, 949–952.
11. F. S. Apple, A. Falahati, P. R. Paulsen, E. A. Miller and S. W. Sharkey, Improved detection of minor ischemic myocardial injury with measurement of serum cardiac troponin I, Clin. Chem., 1997, 43, 2047–2051.
12. F. S. Apple, L. A. Pearce, S. W. Smith, J. M. Kaczmarek and M. A. M. Murakami, Role of monitoring changes in sensitive cardiac troponin I assay results for early diagnosis of myocardial infarction and prediction of risk of adverse events, Clin. Chem., 2009, 55, 930–937.
13. L. Jacquet, P. Noirhomme, G. E. ElKhoury, M. Goenen, M. Philippe, J. Col and R. Dion, Cardiac troponin I as an early marker of myocardial damage after coronary bypass surgery, Eur. J. Cardio-thoracic Surg., 1998, 13, 378–384.
14. I. B. Mănescu, K. Pál, S. Lupu and M. Dobreanu, Conventional biomarkers for predicting clinical outcomes in patients with heart disease, Life, 2022, 12, 2112.
15. A. Sheriff, R. Schindler, B. Vogt, H. Abdel-Aty, J. K. Unger, C. Bock, F. Gebauer, A. Slagman, T. Jerichow, D. Mans, G. Yapici, G. Janelt, M. Schröder, R. Kunze and M. Möckel, Selective apheresis of C–reactive protein: A new therapeutic option in myocardial Infraction?, J. Clin. Apher., 2014, 30, 15–21.
16. J. Simes, K. P. Robledo, H. D. White, D. Espinoza, R. A. Stewart, D. R. Sullivan, T. Zeller, W. Hague, P. J. Nestel, P. P. Glasziou, A. C. Keech, J. Elliott, S. Blankenberg and A. M. Tonkin, D-dimer predicts long-term cause-specific mortality, cardiovascular events, and cancer in patients with stable coronary heart disease LIPID study, Circulation, 2018, 138, 712–723.
17. M. Panteghini, F. S. Apple, R. H. Christenson, F. Dati, J. Mair and A. H. Wu, Use of Biochemical Markers in Acute Coronary Syndromes. IFCC Scientific Division, Committee on Standardization of Markers of Cardiac Damage, Clin. Chem. Lab. Med., 1999, 37, 687–693.
18. J. P. Chapelle, Cardiac troponin I and troponin T: Recent players in the field of myocardial markers, Clin. Chem. Lab. Med., 1999, 37, 11–20.
19. J. H. Keffer, Myocardial Markers of Injury: Evolution and Insights, Am. J. Clin. Pathol., 1996, 105, 305–320.
20. S. Madhurantakam, B. E. David, A. Naqvi, Z. J. Lee, J. T. Abraham, T. S. Vankamamidi and S. Prasad, Advancements in electrochemical immunosensors towards point-of-care detection of cardiac biomarkers, Anal. Methods, 2024, 16, 6615–6633.
21. Z. Beiram Zadeh, S. M. Hosseini, J. Mohammadnejad, M. Tanhaei and S. Ramakrishna, Label-Free Detection of Cardiac Biomarkers: A Review on Microfluidic Electrochemical Biosensors, ACS Appl. Bio Mater., 2023, 6, 2622–2635.
22. S. H. Lee, J. H. Back, H. J. Joo, D. S. Lim, J. E. Lee and H. J. Lee, Simultaneous detection method for two cardiac disease protein biomarkers on a single chip modified with mixed aptamers using surface plasmon resonance, Talanta, 2023, 267, 125232.
23. T. Vairaperumal, C. C. Huang and P. Y. Liu, Optical Nanobiosensor-Based Point-of-Care Testing for Cardiovascular Disease Biomarkers, ACS Appl. Bio Mater., 2023, 6, 2591–2613.
24. P. R. Li, S. Kiran Boilla, C. H. Wang, P. C. Lin, C. N. Kuo, T. H. Tsai and G. BinLee, A self-driven, microfluidic, integrated-circuit biosensing chip for detecting four cardiovascular disease biomarkers, Biosens. Bioelectron., 2023, 249, 115931.
25. Y. Li, X. Liu, J. Guo, Y. Zhang, J. Guo, X. Wu, B. Wang and X. Ma, Simultaneous detection of inflammatory biomarkers by sers nanotag-based lateral flow assay with portable cloud raman spectrometer, Nanomaterials, 2021, 11, 1–14.
26. D. Tsounidi, G. Koukouvinos, V. Christianidis, E. Legaki, V. Giogli, K. Panagiotopoulou, S. Taka, Z. Ekaterinidi, S. Kakabakos, I. Raptis and P. Petrou, Development of a point-of-care system based on white light reflectance spectroscopy: Application in crp determination, Biosensors, 2021, 11, 1–14.
27. H. L. Liu, Y. T. Tseng, M. C. Lai and L. K. Chau, Ultrasensitive and Rapid Detection of N-Terminal Pro-B-Type Natriuretic Peptide (NT-proBNP) Using Fiber Optic Nanogold-Linked Immunosorbent Assay, Biosensors, 2022, 12, 746.
28. L. Huang, Y. Zhang, E. Su, Y. Liu, Y. Deng, L. Jin, Z. Chen, S. Li, Y. Zhao and N. He, Eight biomarkers on a novel strip for early diagnosis of acute myocardial infarction, Nanoscale Adv., 2020, 2, 1138–1143.
29. S. Natarajan and J. Joseph, A novel time-resolved fluorescent lateral flow immunoassay for quantitative detection of the trauma brain injury biomarker-glial fibrillary acidic protein, Sens. Diagn., 2022, 1, 193–197.
30. N. A. Byzova, Y. Y. Vengerov, S. G. Voloshchuk, A. V. Zherdev and B. B. Dzantiev, Development of a lateral flow highway: Ultra-rapid multitracking immunosensor for cardiac markers, Sensors, 2019, 19, 5494.
31. D. Beduk, T. Beduk, A. Ait Lahcen, V. Mani, E. Guler Celik, G. Iskenderoglu, F. Demirci, S. Turhan, O. Ozdogan, S. Ozgur, T. Goksel, K. Turhan, K. N. Salama and S. Timur, Multiplexed aptasensor for detection of acute myocardial infraction (AMI) biomarkers, Sens. Diagn., 2024, 3, 1020–1027.
32. C.-H. Ko and M.-F. Ricky-Lee, Design and fabrication of a microspectrometer based on silicon concave micrograting, Opt. Eng., 2011, 50, 084401.
33. C.-H. K. Ko, C. Hong and W.-H. Chiu, in Next-Generation Spectroscopic Technologies XII, SPIE, 2019, vol. 10983, p.1098303.
34. L. S. Ardekani and P. W. Thulstrup, Gold Nanoparticle-Mediated Lateral Flow Assays for Detection of Host Antibodies and COVID-19 Proteins, Nanomaterials, 2022, 12, 1–17.
35. C. H. Ko, A. B. Tadesse and A. C. Kabiso, Spectrochip-based Calibration Curve Modeling (CCM) for Rapid and Accurate Multiple Analytes Quantification in Urinalysis, Heliyon, 2024, 10, e37722.
36. E. van den Heuvel and Z. Zhan, Myths About Linear and Monotonic Associations: Pearson's r, Spearman's ρ, and Kendall's τ, Am. Stat., 2022, 76, 44–52.
37. D. Tsikas, Mass Spectrometry-Based Evaluation of the Bland–Altman Approach: Review, Discussion, and Proposal, Molecules, 2023, 28, 4905.
38. D. Quesada-González, C. Stefani, I. González, A. dela Escosura-Muñiz, N. Domingo, P. Mutjé and A. Merkoçi, Signal enhancement on gold nanoparticle-based lateral flow tests using cellulose nanofibers, Biosens. Bioelectron., 2019, 141, 111407.
39. X. Chen, L. Ding, X. Huang and Y. Xiong, Tailoring noble metal nanoparticle designs to enable sensitive lateral flow immunoassay, Theranostics, 2022, 12, 574–602.
40. H. Ye, Y. Liu, L. Zhan, Y. Liu and Z. Qin, Signal amplification and quantification on lateral flow assays by laser excitation of plasmonic nanomaterials, Theranostics, 2020, 10, 4359–4373.
41. J. Hu, L. Wang, F. Li, Y. L. Han, M. Lin, T. J. Lu and F. Xu, Oligonucleotide-linked gold nanoparticle aggregates for enhanced sensitivity in lateral flow assays, Lab Chip, 2013, 13, 4352–4357.
42. W. B. Wu, L. Zhan, J. Wang and C. Z. Huang, A plasmon resonance light scattering assay of glucose based on the formation of gold nanoparticles, Anal. Methods, 2014, 6, 3779–3783.
43. C. H. Ko, W. Y. Kong, A. C. Kabiso, W. H. Chiu, A. B. Tadesse, C. Hong, C. F. Wu and H. H. Lin, Portable SpectroChip-Based Immunoassay Platform for Rapid and Accurate Melamine Quantification in Urine Samples, Toxics, 2024, 12, 870.
44. W. H. Chiu, W. Y. Kong, Y. J. Su, J. W. Wen, C. M. Tsai, C. Hong, P. Y. Chen and C. H. Ko, A Faster, Novel Technique to Detect COVID-19 Neutralizing Antibodies, Med. Sci. Monit., 2022, 28, 1–9.
45. P. Y. Chen, C. H. Ko, C. J. Wang, C. W. Chen, W. H. Chiu, C. Hong, H. M. Cheng and I. J. Wang, The early detection of immunoglobulins via optical-based lateral flow immunoassay platform in COVID-19 pandemic, PLoS One, 2021, 16, 1–13.
46. D. J. W. Lee and T.-C. Aw, Population differences and high-sensitivity troponin values: a narrative review, J. Lab. Precis. Med., 2023, 8, 14.
47. J. Wu, Y. Hua, Y. Ge, K. Chen, S. Chen, J. Yang and H. Yuan, Clinical performance validation and four diagnostic strategy assessments of high-sensitivity troponin I assays, Sci. Rep., 2025, 15, 14442.
48. T. Kristian, A. Joseph S, J. Allan S, C. Bernard R, B. Jeroen J, M. David A and W. Harvey D, Fourth Universal Definition of Myocardial Infarction, JACC, 2018, 72, 2231–2264.
49. N. Rifai, A. R. Horvath and C. Wittwer, Tietz Textbook of Clinical Chemistry and Molecular Diagnostics, Elsevier, 2018.
50. H.-P. Brunner-La Rocca and S. Sanders-van Wijk, Natriuretic Peptides in Chronic Heart Failure., Card. Fail. Rev., 2019, 5, 44–49.
51. P. Welsh, R. T. Campbell, L. Mooney, D. M. Kimenai, C. Hayward, A. Campbell, D. Porteous, N. L. Mills, N. N. Lang, M. C. Petrie, J. L. Januzzi, J. J. V. McMurray and N. Sattar, Reference Ranges for NT-proBNP (N-Terminal Pro-B-Type Natriuretic Peptide) and Risk Factors for Higher NT-proBNP Concentrations in a Large General Population Cohort, Circ. Heart Fail., 2022, 15, e009427.
52. P. Feng, J. Qin, Z. Chai, Y. Zhang, M. Zhao, L. Liu, L. Zhang, Y. Chen and Y. Xiong, Analytical and clinical performance evaluation of a new NT-proBNP assay, BMC Cardiovasc. Disord., 2024, 24, 341.
53. Z. Cao, Y. Jia and B. Zhu, BNP and NT-proBNP as Diagnostic Biomarkers for Cardiac Dysfunction in Both Clinical and Forensic Medicine, 2019,  DOI:10.3390/ijms20081820.
54. A. Mouzarou, N. Hadjigeorgiou, D. Melanarkiti and T. E. Plakomyti, The Role of NT-proBNP Levels in the Diagnosis of Hypertensive Heart Disease, 2025,  DOI:10.3390/diagnostics15010113.
55. T. Bertsch, W. Behringer, S. Blaschke, R. Body, S. Davidson, M. Müller-Olling, G. Guo, A. Rieger, A. Wahl, D. Horner, Y. Sun, L. Turnes, U. Sonner and M. Hoffmann, Deep vein thrombosis and pulmonary embolism: a prospective, observational study to evaluate diagnostic performance of the Tina-quant D-Dimer Gen.2 assay, Front. Cardiovasc. Med., 2023, 10, 1–9.
56. L. Barrett, T. Jones and D. Horner, The application of an age adjusted D-dimer threshold to rule out suspected venous thromboembolism (VTE) in an emergency department setting: a retrospective diagnostic cohort study, BMC Emerg. Med., 2022, 22, 186.
57. M. Franchini, D. Focosi, M. P. Pezzo and P. M. Mannucci, How we manage a high D-dimer, Haematologica, 2024, 109, 1035–1045.
58. J. F. Thachil, J. Emmanuel and L. Giuseppe, Giuseppe, D-dimers—”Normal” Levels versus Elevated Levels Due to a Range of Conditions, Including “D-dimeritis”, Inflammation, Thromboembolism, Disseminated Intravascular Coagulation, and COVID-19, Semin. Thromb. Hemostasis, 2022, 48, 672–679.
59. M. Alshalhoub, F. Alhusain, F. Alsulaiman, A. Alturki, S. Aldayel and M. Alsalamah, Clinical significance of elevated D-dimer in emergency department patients: a retrospective single-center analysis, Int. J. Emerg. Med., 2024, 17, 47.
60. P. L. Brill-Edwards Agnes, D-Dimer, Testing in the Diagnosis of Acute Venous Thromboembolism, Thromb. Hemostasis, 1999, 82, 688–694.
61. M. Akhter, J. Kline, B. Bhattarai, M. Courtney and C. Kabrhel, Ruling out pulmonary embolism in patients with high pretest probability, West. J. Emerg. Med., 2018, 19, 487–493.
62. G. D. M. Reber Philippe, D-Dimer, Assays for the Exclusion of Venous Thromboembolism, Semin. Thromb. Hemostasis, 2000, 26, 619–624.
63. E. Amezcua-Castillo, H. González-Pacheco, A. Sáenz-San Martín, P. Méndez-Ocampo, I. Gutierrez-Moctezuma, F. Massó, D. Sierra-Lara, R. Springall, E. Rodríguez, A. Arias-Mendoza and L. M. Amezcua-Guerra, C-Reactive Protein: The Quintessential Marker of Systemic Inflammation in Coronary Artery Disease-Advancing toward Precision Medicine, Biomedicines, 2023, 11, 2444.
64. P. M. Ridker, Clinical Application of C-Reactive Protein for Cardiovascular Disease Detection and Prevention, Circulation, 2003, 107, 363–369.
65. T. A. Pearson, G. A. Mensah, R. W. Alexander, J. L. Anderson, R. O. Cannon, M. Criqui, Y. Y. Fadl, S. P. Fortmann, Y. Hong, G. L. Myers, N. Rifai, S. C. Smith, K. Taubert, R. P. Tracy and F. Vinicor, Markers of Inflammation and Cardiovascular Disease, Circulation, 2003, 107, 499–511.
66. I. Kushner, D. Rzewnicki and D. Samols, What Does Minor Elevation of C-Reactive Protein Signify?, Am. J. Med., 2006, 119, 166.
67. K. Amit, M. Darren K, M. Sabina A, S. Harold G, D. Sandeep R, V. Wanpen, W. Frank H, G. Scott M and de L. James A, Race and Gender Differences in C-Reactive Protein Levels, JACC, 2005, 46, 464–469.
68. K. S. Chew, What’s new in Emergencies Trauma and Shock? C-reactive protein as a potential clinical biomarker for influenza infection: More questions than answers, J. Emerg. Trauma Shock, 2012, 5, 115–117.

---