Estimating Work Engagement with Wrist-Worn Heart Rate Sensors

This study aims to estimate the work engagement (WE) of office workers using biological data related to their daily activities obtained from wearable devices, with the goal of providing appropriate mental and physical health support to each individual. We collected daily heart rate data from wearable devices worn by 60 office workers in five Japanese companies for 2-3 weeks. Their daily WE was measured using the state-of the-art utrecht work engagement scale (UWES) questionnaire. We performed two types of analysis on the collected data using machine learning methods. First, the classification of binary WE levels (high or low), which showed a leave-one-person-out (LOPO) cross-validation F1 value of 0.522. Second, we classified whether WE decreased compared to the previous day, which showed a LOPO cross-validation F1 value of 0.663.