Towards intelligent health monitoring: A CNN-driven ultrasensitive wearable pressure sensor featuring a bio-inspired dual-layer microstructure

The growing need for health monitoring, intelligent recognition, and safety assurance has driven significant interest in wearable sensing technologies. Owing to their compact size, wearability, and real-time monitoring capability, wearable sensors are increasingly used in physiological signal detection, intelligent recognition systems, and human–machine interfaces. In this work, we present a novel wearable flexible pressure sensor (WFPS) inspired by the neural tactile sensing mechanism of human skin. The sensor features a bio-inspired dual-layer microstructure design, which enhances its performance compared to conventional single-layer architectures. Specifically, the dual-layer WFPS exhibits a 57% improvement in output performance, with a sensitivity of 6.13 V/kPa—a 48% increase over single-layer devices. It reliably detects subtle pressure variations down to 1.4 Pa, shows a fast response time of 6 ms, and maintains stable operation over 2500 loading–unloading cycles. These superior characteristics enable the WFPS to monitor diverse physiological signals such as carotid pulse, laryngeal vibration, and limb movement. Furthermore, we demonstrate its utility in intelligent recognition systems through a convolutional neural network (CNN)-based handwritten digit recognition platform. Using sensor data acquired via a signal collection system, the platform achieves a recognition accuracy of 98.3% across ten digit classes. Additionally, a driver fatigue monitoring system integrating the WFPS with a hybrid CNN-LSTM algorithm is developed, enabling real-time fatigue detection with an accuracy of 96.7%. The promising performance of the WFPS in health monitoring and intelligent recognition underscores its potential for advancing future wearable technologies and smart sensor systems.