Development of an intelligent wearable system for monitoring physiological parameters and stress based on microcontrollers and modern data processing algorithms

The article presents the development of an intelligent wearable system for continuous monitoring of physiological parameters and early detection of stress conditions using modern microcontrollers and machine learning algorithms. The system implements a multisensor approach that includes the registration of electrodermal activity (EDA), heart rate (HR), body temperature, and accelerometer data. For automatic stress level classification, several machine learning models were compared using the main metrics (accuracy, precision, recall, F1-score). The best accuracy was achieved using the Random Forest model: accuracy 91%, F1-score 0.90. The device's experimental autonomy was 36 hours in monitoring mode during testing on 25 volunteers. The article presents the results of experimental tests that confirm the significance of physiological indicators for stress detection, and describes the architecture of energy-saving algorithms that allow for increasing the device's autonomy. The developed system provides personalized monitoring of the user's condition and can be used for stress disorder prevention, remote medical monitoring, and integration into smart healthcare solutions.