AI-driven smart mask-based respiratory monitoring system utilizing a high-performance and biocompatible PVA/PVP/graphene composite thermoelectric film

Respiratory monitoring, as a fundamental approach to assessing vital signs and health conditions, plays an irreplaceable role in early disease warning and personalized healthcare. However, conventional respiratory monitoring technologies are often limited by device rigidity and poor portability, severely restricting their use in daily monitoring. To address this issue, we propose a smart mask-based respiratory monitoring system based on a high-performance and biocompatible composite thermoelectric film (CTF). The CTF comprises electrospun polyvinyl alcohol nanofiber membrane as a flexible substrate, and screen-printed polyvinyl pyrrolidone/graphene-based conductive ink as the thermoelectric layer. Experimental results indicate that the CTF-based thermoelectric device demonstrates superior performance, characterized by a high Seebeck coefficient (20.6 μV/K), a rapid response time (0.4 s), an ultra-low detection limit (0.5 K), excellent long-term stability (200 cycles), and notable biocompatibility (cell viability: 97.5 %). In addition, the CTF exhibits excellent mechanical properties and enables non-contact sensing. By integrating CTF into a face mask and leveraging embedded systems and wireless transmission technology, respiratory signals can be effectively collected. A custom-designed convolutional neural network enables the classification of five different abnormal breathing patterns, achieving an accuracy of 97.4 %. This work offers an effective strategy for respiratory disease monitoring and holds significant potential for daily health tracking and clinical applications.