Self-powered porous polymer sensors with high sensitivity for machine learning-assisted motion and rehabilitation monitoring

Abstract

Muscle contraction and relaxation inherently contains valuable information crucial for monitoring physical states, rehabilitation, and injury prevention. However, the majority of flexible wearable devices struggle to offer customizable sensors with high sensitivity, durability, and stability, resulting in suboptimal biofeedback performance. Herein, the machine learning-assisted smart motion and rehabilitation monitoring system (SMRMS) is demonstrated for full-body motion recognition and rehabilitation assessment using a designed porous triboelectric nanogenerator (TENG) array. A theoretical model of porous materials is built to demonstrate the effect mechanism of porosity on TENG output. Through pore design on the tribolayer, theoretical simulation and experimental are performed to determine the key features of porous TENG sensors with high sensitivity (1.76 kPa−1), fast response time (50 ms) and high durability (over 100,000 cycles). A ring-shaped TENG (RS-TENG) is fabricated based on the porous TENG sensor, enabling real-time recording of leg/arm force without additional attachment. By integrating the RS-TENG array with a multichannel signal acquisition system, an SMRMS is designed to capture motion information during human subjects’ exercise and rehabilitation training. These data are then fused for machine learning analytics, resulting in significantly improved accuracy in motion pattern recognition (98.75 %) and rehabilitation monitoring (100 %). The simple, precise and durable porous sensors could help mitigate the risk of excessive exercise-induced muscle injuries, expanding self-powered wearable functionalities and adaptabilities.