AI-Driven Self-Powered Wearable Sensors for Gait Detection

Abstract

Abstract Against the backdrop of rapid advancements in wearable biosensing technology, hydrogel-based triboelectric nanogenerators (TENGs) have garnered significant attention due to their flexibility and self-powered sensing capabilities. However, TENGs still face numerous obstacles in terms of sensitivity, long-term stability, and data processing. This paper describes a TENG named DHXN-TENG, which uses a hydrogel named DHXN as its substrate and features a microneedle structure on its upper surface. The DHXN hydrogel exhibits outstanding performance, including high sensitivity (gauge factor (GF): 5.46, strain range: 510%), fast response time (60 ms), fatigue resistance, and good biocompatibility. The DHXN-TENG based on the DHXN hydrogel has an extremely short response time (30 ms), overcoming the limitation of existing self-powered sensors that struggle to acquire signals in real time due to dynamic response hysteresis. At the same time, the DHXN-TENG has a maximum power output of 8.38 W m–2 and an extremely high surface charge density, fundamentally eliminating the risk of battery leakage. Furthermore, the DHXN-TENG demonstrates excellent output stability in environments with fluctuating temperature and humidity while also offering reliable long-term durability. In this study, leveraging the exceptional sensing performance of the DHXN-TENG and integrating wireless Bluetooth technology, we developed a wireless remote intelligent monitoring system for foot movement patterns. The system incorporates deep learning algorithms based on convolutional neural networks for gait recognition and rehabilitation assistance. The DHXN-TENG demonstrates significant potential in the fields of medical rehabilitation and human–computer interaction technology.