The article discusses a study that used machine learning to identify stress in college students using data collected from wearable sensors. The sensors, worn on the wrist, gathered data on heart rate and hand acceleration. The study involved 54 students who used the sensors and a mobile health app for 40 days. The application also allowed users to self-report stress. The most reliable model for identifying stress was the XGBoost method, with an accuracy of 84.5%. The findings suggest that patterns in physiological reaction to stress can be identified using smartwatch sensors, informing the design of future stress detection tools.
Publication date: 21 Sep 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2309.11097
