Unobtrusive stress detection using wearables: application and challenges in a university setting

Introduction In theory, wearable physiological sensing devices offer an opportunity for institutions to monitor and manage the health and well-being of a group of people. For instance, schools or universities could leverage these devices to track rising stress levels or detect signs of illness among students. Advances in sensing accuracy and utility design in wearables might make this feasible; however, real-world adoption faces challenges, as users often fail to wear or use these devices consistently and correctly. Additionally, institutional monitoring raises privacy concerns. Methods In this study, we analyze real-world data from a cohort of 103 Japanese university students to identify periods of cyclical stress while ensuring individual privacy through aggregation. We identify potential stress patterns by observing elevated waking heart rate (HR) and maximum waking HR, supported by related metrics such as sleep HR, sleep heart rate variability (HRV), activity patterns, and sleep phases. Results The physiological changes align with significant academic and societal events, indicating a strong link to stress. Discussion Our findings demonstrate the potential of consumer wearables to detect collective changes in stress biomarkers within a cohort using in-the-wild data, i.e., data that is noisy and has gaps. Furthermore, we explore how universities could implement such monitoring in practice, highlighting both the potential benefits and challenges of real-world application.