Overexertion-related construction activities are identified as a leading cause of work-related musculoskeletal disorders (WMSDs) among construction workers. However, few studies have focused on the automated recognition of overexertion-related construction workers' activities as well as assessing ergonomic risk levels, which may help to minimize WMSDs. Therefore, this study examined the feasibility of using acceleration and foot plantar pressure distribution data captured by a wearable insole pressure system for automated recognition of overexertion-related construction workers' activities and for assessing ergonomic risk levels. The proposed approach was tested by simulating overexertion-related construction activities in a laboratory setting. The classification accuracy of five types of supervised machine learning classifiers was evaluated with different window sizes to investigate classification performance and further estimate physical intensity, activity duration, and frequency information. Cross-validation results showed that the Random Forest classifier with a 2.56-s window size achieved the best classification accuracy of 98.3% and a sensitivity of more than 95.8% for each category of activities using the best features of combined data set. Furthermore, the estimation of corresponding ergonomic risk levels was within the same level of risk. The findings may help to develop a noninvasive wearable insole pressure system for the continuous monitoring and automated activity recognition, which could assist researchers and safety managers in identifying and assessing overexertion-related construction activities for minimizing the development of WMSDs' risks among construction workers.
|Journal||Journal of Construction Engineering and Management|
|Publication status||Published - 1 Jul 2020|