AI‐enabled biomimetic flexible pressure sensors for high‐precision gait monitoring and rehabilitation assessment

Abstract Flexible pressure sensors, owing to their excellent comfort, flexibility, and adaptability, have emerged as promising candidates for applications in smart healthcare and rehabilitation training. However, challenges such as suboptimal sensing performance, insufficient stability, integration difficulties, and the real‐time interpretation of pathological gait characteristics still severely constrain their reliability and practicality in wearable applications. Here, we report multilayer microstructured capacitive sensors based on hierarchical polymer composites. These sensors demonstrate ultrahigh sensitivity (24.3% kPa −1 ), an ultralow detection limit (0.49 Pa), a broad pressure range (up to 500 kPa), and outstanding mechanical stability over 6000 loading‐unloading cycles. Finite element simulations reveal that both the TPU/Ag electrospun nanofiber film and the porous dielectric layer undergo compression under loading, thereby enhancing the air‐gap effect and dielectric properties, which collectively boost sensing sensitivity. The simulations further validate the working mechanism of the sensors under different pressure conditions, in good agreement with experimental results. Moreover, the sensors exhibit excellent stability even under extreme pressures (e.g., during automobile driving), and can be applied in human joint motion monitoring and encrypted communication recognition. By embedding artificial intelligence algorithms, spatiotemporal pressure data collected by sensor arrays are decoded and wirelessly transmitted to a mobile application. These models achieve 98.96% accuracy in detecting gait recovery stages of patients after knee surgery and 99.44% accuracy in classifying six rehabilitation training gait patterns, thereby laying a solid foundation for the development of next‐generation high‐performance intelligent insoles based on flexible pressure sensors.