Use of Wearable-Generated Real-World Data to Objectively Identify Occupational Stressors

OBJECTIVES: This study aims to identify a meaningful combination of heart rate variability (HRV)-related indices (HRVIs) to investigate physiological responses to occupational stressors such as high workload with the potential to predict emotional distress or even burnout at early stages using AI-based approaches.

METHODS: Wearable-based HRV measurement is performed using optical heart rate sensors, providing pulse-to-pulse intervals (PPI) of volunteers within the office context. Based on PPI sequences several HRVIs (e.g., Baevsky stress index (SI) or heart rate) were calculated. k-means clustering was used to partition data into a pre-defined number of clusters to describe individual states. Clusters were analyzed over working hours, days and weeks to draw conclusions about particularly stressful periods at work. In a unique approach, individual mood tracking, using the circumplex model-based Self-Assessment Manikins (Bradley & Lang, 1994) describing valence and arousal, was used to map the identified clusters to different emotional states. Application- and time-dependent HRVI fluctuations were determined by combining SI with synchronized information on the program used, obtained from a software tracking tool.

RESULTS: Seven HRVIs were considered meaningful as well as computationally effective for predicting emotional states. Data analysis showed that emotional states varied by day, weekday and week, and could be attributed to work activities, appointments and recovery time, indicating the most stressful periods. Moreover, HRV values changed based on the usage of different software programs, with some applications displaying above-average SI values, which may indicate stress or mental strain.

CONCLUSIONS: The combination of HRVIs with application usage data provides valuable insights into stress and workload patterns in digital work environments. Future work will consider the use of mouse and keyboard dynamics data as additional source for better mapping of emotional states. These approaches aim to improve the understanding of the dominant emotions during the working day to assess job satisfaction.