Wonil Lee

Project title: Occupational fatigue prediction for entry-level construction workers in material handling activities using wearable sensors

Degree: MS (Thesis) | Program: Occupational & Environmental Exposure Sciences (OEES) | Project type: Thesis/Dissertation
Completed in: 2018 | Faculty advisor: Edmund Y. W. Seto


Research on the measurement and prediction of occupational fatigue of construction workers using wearable sensors has been carried out using different types of sensor technologies and measurement variables. Previous studies have demonstrated promising results using wearable sensors for fatigue prediction, suggesting that they may have practical use as tools for the prevention of fatigue management on worksites. However, there are no clear guidelines on the type of sensors to use in fatigue management or on the relevant sensed variables. Moreover, the collection and processing of wearable sensor data should be sufficiently simple for safety professionals to use in practice. The current study aimed to address these challenges by using several of the most active wearable sensor technologies in occupational fatigue research—actigraphy and electrocardiogram (ECG) sensors—to obtain different variables from participants performing simulated construction tasks in laboratory conditions. A total of 22 participants participated in the experiment. Of these, 19 were exposed to different task workloads and completed four repetition measurements, while three participants only completed one or two experiment session(s). A total of 80 observations obtained by the experiment were used for the analysis. Stepwise logistic regression was used to identify the most appropriate and parsimonious fatigue prediction model. Among the different variables collected, heart rate variability (HRV) measurements, standard deviation of NN intervals (SDNN) and power in the low frequency range (LF) were found to be useful in predicting fatigue. Both the fast Fourier transform (FFT) and the autoregressive (AR) analysis in the frequency domain analysis methods were employed. Log transformed LF obtained by AR analysis method was found to be more suitable for daily management of worker’s fatigue, while the SDNN was useful in weekly fatigue management. This study contributes to the body of knowledge on the use of wearable technology for the management of fatigue among construction workers. URI