A Machine-Learning-Assisted Wireless Battery-Free Stress Monitoring Wearable System

Traditional stress assessment methods, including questionnaires and laboratory assays, suffer from limitations such as non-quantitative nature, lack of standardization, and impracticality for continuous monitoring. To address these challenges, we have developed a wireless battery-free stress monitoring wearable system equipped with machine learning (ML) using Ag electrodes as epidermal sensors for galvanic skin response (GSR) signal monitoring. These ultrathin, stretchable electrodes are fabricated using cost-efficient screen-printing methods, ensuring low contact impedance, high signal-to-noise ratio, and excellent mechanical stability. For signal collection, we employed a battery-free wireless readout system. The system utilizes Near-Field Communication (NFC) technology for wireless power and data transmission, eliminating the need for bulky batteries and enhancing wearability and comfort. The collected GSR data is transmitted to a mobile phone for real-time processing and analysis. Machine learning models were employed to realize stress event classification, accurately distinguishing between stress and rest states based on the GSR signals. We conducted in-situ experiments to evaluate the sensor's capability to detect stressful events during a Stroop test. The results demonstrated the system's effectiveness in providing continuous, real-time stress monitoring in everyday settings. Our innovative wearable device represents a significant advancement in stress assessment technology, offering a practical and efficient solution for continuous health monitoring.