Abstract
Work related activities that led to overexertion are a major cause of work-related
musculoskeletal disorders (WMSDs) among construction workers. However, existing risk
assessment methods (e.g., self-reported and observational-based methods) have failed
to fully recognize these activities and assess the corresponding risk level exposure to
mitigate WMSDs. This study examines the feasibility of using acceleration and foot
plantar pressure distribution data captured by a wearable insole pressure system for
automated assessment of construction workers’ activities and overexertion risk levels.
The accuracy of five types of supervised machine learning classifiers was evaluated with
different window sizes to investigate individual participant performance and further
estimate physical intensity, activity duration and frequency information. The results
showed that the Random Forest classifier with 2.56s window size achieved the best
classification accuracy of 94.5% and 94.3% and a sensitivity of more than 90.1% and
88.4% for each category of activities. Overall, the proposed approach provides a noninvasive method and objective assessment of ergonomic risk level based on
acceleration and foot plantar pressure distribution data captured by a wearable insole
pressure system which could help other researchers and safety managers to: understand
the level of workers’ risks; and provide an effective intervention to mitigate the risk of
developing WMSDs among construction workers.
musculoskeletal disorders (WMSDs) among construction workers. However, existing risk
assessment methods (e.g., self-reported and observational-based methods) have failed
to fully recognize these activities and assess the corresponding risk level exposure to
mitigate WMSDs. This study examines the feasibility of using acceleration and foot
plantar pressure distribution data captured by a wearable insole pressure system for
automated assessment of construction workers’ activities and overexertion risk levels.
The accuracy of five types of supervised machine learning classifiers was evaluated with
different window sizes to investigate individual participant performance and further
estimate physical intensity, activity duration and frequency information. The results
showed that the Random Forest classifier with 2.56s window size achieved the best
classification accuracy of 94.5% and 94.3% and a sensitivity of more than 90.1% and
88.4% for each category of activities. Overall, the proposed approach provides a noninvasive method and objective assessment of ergonomic risk level based on
acceleration and foot plantar pressure distribution data captured by a wearable insole
pressure system which could help other researchers and safety managers to: understand
the level of workers’ risks; and provide an effective intervention to mitigate the risk of
developing WMSDs among construction workers.
| Original language | English |
|---|---|
| Title of host publication | 8th West Africa Built Environment Research (WABER) Conference |
| Pages | 788-796 |
| ISBN (Electronic) | 978 – 9988 – 2– 6010 – 1 |
| DOIs | |
| Publication status | Published - 7 Aug 2019 |
| Event | 8th West Africa Built Environment Research (WABER) Conference - Accra, Ghana Duration: 5 Aug 2019 → 7 Aug 2019 |
Conference
| Conference | 8th West Africa Built Environment Research (WABER) Conference |
|---|---|
| Country/Territory | Ghana |
| City | Accra |
| Period | 5/08/19 → 7/08/19 |