IoT Wearable Devices to Detect and Monitor Mental Health and Stress

With personalized healthcare services increasing in demand and with the perception that the incidences of mental diseases are on the upward path, different types of interests have been implemented in the development of wearable IoT-based devices for mental health monitoring. These devices provide real-time insights into a person's wellbeing by capturing behavioral and physiological indicators that include heart rate, skin conductance, body temperature, and activity level. A wearable system was developed in this paper based on sensors comprising GSR, ECG (AD8232), IR temperature sensor (MLX90614), and accelerometer, with data processing being performed via the ESP32 microcontroller. The sensor data were sent to some cloud platforms via MQTT and Wi-Fi protocols and analyzed by a pre-trained XGBoost Machine Learning model. The system was trained and tested with the Stress-Lysis dataset, which contains physiological and behavioral features. Experimental evaluation proved the model's F1-score of 0.90, thus beating Logistic Regression, Random Forest, and SVM classifiers. The system performed with good real-time stress classification into LOW, MEDIUM, and HIGH levels, thus delivering alerts with great timeliness through the app and email.