Machine Learning-Enabled Planar Interdigitated Thermogalvanic Hydrogel for Synergistic Thermal and Strain Sensing

Abstract

Thermal and strain recognition is a promising modality for wearable electronics and human-machine interfaces, taking advantage of its natural, efficient, and userdistinctive characteristic. However, current sensing materials are typically limited by structural rigidity, functional constraints, and significant external power dependency, hindering their practical applications. This work proposes a unique strain-thermal sensing mechanism based on a planar interdigitated thermogalvanic hydrogel (PITH), integrating MXene interdigitated electrodes with a polyacrylamide (PAAm)-Fe(CN)₆³⁻/⁴⁻ hydrogel. Operating under a temperature gradient, this self-powered device responds to hand-writing motions, thereby enabling real-time capture of subtle writing dynamics. Distinct from resistive strain-based designs, the PITH employs a strain-thermal coupling mechanism, where mechanical strain modulates the hydrogel's redox kinetics, thereby influencing thermoelectric voltage output and establishing a direct strain-voltage correlation. Through random forest algorithm processing dynamic 2 voltage signals, discriminative feature is extracted, achieving high recognition accuracy (99.09% for letters, 97.85% for digits) under entirely passive operation. This study presents a structurally integrated, fully self-powered, and signal-programmable flexible handwriting recognition device, which is important for developing next-generation low-power wearable HMI systems.