A data-driven approach to understand factors contributing to exoskeleton use-intention in construction

Abstract

Work-related musculoskeletal disorders (WMSDs) remain an important heath concern for construction workers. Occupational exoskeletons (EXOs) are a new ergonomic intervention to control WMSD risk, yet their adoption has been low in construction. We explored contributing factors to EXO use-intention, by building a decision tree to predict the intention to try an exoskeleton using responses to an online survey. Variable selection and hyperparameter tuning were used respectively to reduce the number of potential predictors, and for a better prediction performance. Performance was assessed using four common metrics. The importance of variables in the final tree was calculated to understand which variable had a greater influence. The final tree had moderate prediction performance. Important variables identified were associated with opinions on EXO use, demographics, job demands, and perceived potential risks. The key influential variables were EXOs becoming standard equipment and fatigue reduction with EXO use. Practical implications of the findings are discussed.