Context-Aware Stress Monitoring using Wearable and Mobile Technologies in Everyday Settings

Background Daily monitoring of stress is a critical component of maintaining optimal physical and mental health. Physiological signals and contextual information have recently emerged as promising indicators for detecting instances of heightened stress. Nonetheless, developing a real-time monitoring system that utilizes both physiological and contextual data to anticipate stress levels in everyday settings while also gathering stress labels from participants represents a significant challenge. Objective We present a monitoring system that objectively tracks daily stress levels by utilizing both physiological and contextual data in a daily-life environment. Additionally, we have integrated a smart labeling approach to optimize the ecological momentary assessment (EMA) collection, which is required for building machine learning models for stress detection. We propose a three-tier Internet-of-Things-based system architecture to address the challenges. Methods A group of university students (n=11) consisting of both males (n=4) and females (n=7) with ages ranging from 18 to 37 years (Mean = 22.91, SD = 5.05) were recruited from the University of California, Irvine. During a period of two weeks, the students wore a smartwatch that continuously monitored their physiology and activity levels. A context-logging application was also installed on their smartphone. They were asked to respond to several EMAs daily through a smart EMA query system. We employed three different machine learning algorithms to evaluate the performance of our system. The mean decrease impurity approach was employed to identify the most significant features. The k-nearest neighbor imputation technique was used to fill out the missing contextual features. Results F1-score is the performance metric used in our study. We utilized a cross-validation technique to accurately estimate the performance of our stress models. We achieved the F1-score of 70% with a Random Forest classifier using both PPG and contextual data, which is considered an acceptable score in models built for everyday settings. Whereas using PPG data alone, the highest F1-score achieved is approximately 56%, emphasizing the significance of incorporating both PPG and contextual data in stress detection tasks. Conclusion We proposed a system for monitoring daily-life stress using both physiological and smartphone data. The system includes a smart query module to capture high-quality labels. This is the first system to employ both physiology and context data for stress monitoring and to include a smart query system for capturing frequent self-reported data throughout the day.