A self-powered flexible piezoelectric sensor patch for deep learning-assisted motion identification and rehabilitation training system

Abstract

Artificial intelligence-assisted wearable devices have attracted great interest in medical treatment and healthcare. However, wearable electronic devices are expensive to manufacture and usually depend on external power supply. Herein, a flexible self-powered piezoelectric sensor patch (SPP) using Polyvinylidene fluoride (PVDF) fibrous film as the functional layer is demonstrated for the assessment and motion identification of wrist joint rehabilitation training. PVDF fibrous film is prepared by a triboelectric nanogenerator (TENG)-driven near-field electrospinning system with a special designed synchronous mechanical switch. The results show that this flexible SPP has a high sensitivity of 0.2768 V KPa−1 at pressures from 1 to 75 kPa. Such excellent flexibility allows us to attach the SPP to the finger as a tactile sensor for rehabilitation assessment of wrist joint flexibility. In addition, long short-term memory network model is used to process the collected data from the SPP for motion identification. The test accuracy of the SPP wrist motion identification reaches 92.6%, which afford a potential way to understand the progress of the rehabilitation training of patients' wrists. Generally, this flexible SPP shows great promise for applications in the fields of motion monitoring, medical diagnosis and rehabilitation training based on artificial intelligence.