BTS(Big data Technology and Strategy)

논문 초록 (Abstract)

Purpose Prediction model for occupational injuries support more proactive, efficient, and effective policies. This study aims to develop a prediction model for occupational injuries severity, classified into 15 disability grades in South Korea, using machine learning techniques on COMWEL big data. The primary goal is to enhance prediction accuracy, providing a advanced policy tool for early intervention and for evidence-based policy operations. Design/methodology/approach The data analyzed in this study comprises 290,157 administrative records of occupational injuries cases collected from 2018 to 2020 by the Korea Workers' Compensation & Welfare Service, based on the ‘Workers’ Compensation Insurance Application Form’ submitted for occupational injuries treatment. Four machine learning models — Decision Tree, FCNN, XGBoost, and LightGBM — were developed and their performances compared to identify the optimal model. Additionally, the Permutation Feature Importance(PFI) method was employed to estimate the relative contribution of each variable to the performance of the predictive model, thereby identifying key variables. Findings The FCNN algorithm achieved the lowest MAE of 0.7276. Key variables for predicting disability grades included severity index, primary disease code, primary disease site, age at the time of the occupational injuries, and type of industry. This highlights the importance of early policy intervention, and of both medical and socioeconomic factors in model prediction. Academic and policy implications were discussed based on these results.