Selection of wearable sensor measurements for monitoring and managing entry-level construction worker fatigue: A logistic regression approach

Purpose: The identification of fatigue status and early intervention to mitigate fatigue can reduce the risk of workplace injuries. Off-the-shelf wearable sensors capable of assessing multiple parameters are available. However, using numerous variables in the fatigue prediction model can elicit data issues. This study aimed at identifying the most relevant variables for measuring occupational fatigue among entry-level construction workers by using common wearable sensor technologies, such as electrocardiogram and actigraphy sensors. Design/methodology/approach: Twenty-two individuals were assigned different task workloads in repeated sessions. Stepwise logistic regression was used to identify the most parsimonious fatigue prediction model. Heart rate variability measurements, standard deviation of NN intervals and power in the low-frequency range (LF) were considered for fatigue prediction. Fast Fourier transform and autoregressive (AR) analysis were employed as frequency domain analysis methods. Findings: The log-transformed LF obtained using AR analysis is preferred for daily fatigue management, whereas the standard deviation of normal-to-normal NN is useful in weekly fatigue management. Research limitations/implications: This study was conducted with entry-level construction workers who are involved in manual material handling activities. The findings of this study are applicable to this group. Originality/value: This is the first study to investigate all major measures obtainable through electrocardiogram and actigraphy among current mainstream wearables for monitoring occupational fatigue in the construction industry. It contributes knowledge on the use of wearable technology for managing occupational fatigue among entry-level construction workers engaged in material handling activities.