Although there have been many studies related to construction site safety that have tried to reduce accidents, no significant improvement has been reported. Because of the complex nature of construction work processes, it is important to have a scenario-based worksite safety management system instead of reports such as safety guidelines and manuals. This study utilizes accumulated construction site accident big data, namely Construction Safety Management Integrated Information (CSI), to establish accident scenarios for different work types. To propose accident occurrence scenarios, the hazard profile managed by CSI and prior research analyses are employed for each work type and cause materials at the construction site. For accident occurrence association rules, we developed a framework based on Work Breakdown Structure–Risk Breakdown Structure (WBS-RBS) for reinforced concrete work, temporary work, and earthwork, considering 25,986 accident cases. Subsequently, association analysis was conducted to derive association rules for each work type. The accident occurrence scenarios were extracted by classifying work types (project type, activity type) and cause materials (object, location). The analysis generated 145 association rules, and 76 association rules for reinforced concrete work, temporary work, and earthwork work were extracted to derive accident scenarios, considering scenarios with high accident frequency. Furthermore, by establishing association rules between work processes, we derived accident types and occurrence rules frequently observed at construction sites. These rules formed the basis for constructing accident occurrence scenarios for each construction type based on WBS-RBS. These findings facilitate the development of appropriate safety management plans and effective accident countermeasures tailored to specific construction types and causes. The developed scenarios will help to improve construction site safety by providing useful information for safety managers and worker training.

The construction industry has persistently high accident rates, and major events continue despite strengthened safety management systems. This study analyzes 19,456 accident records from the national Construction Safety Management Integrated Information (CSI) system and applies a Light Gradient Boosting Machine (LightGBM) model to predict fatal versus injury outcomes. SHAP was used to identify influential factors and quantify each variable’s contribution. Fatal events represented about 5% of cases, reflecting substantial class imbalance. To address this, three oversampling methods—SMOTE, Borderline-SMOTE, and ADASYN—were tested. The ADASYN model showed the best performance (F1-score = 0.905, AUC = 0.879) and was selected as the final model. Oversampling was applied exclusively to the training folds during stratified 10-fold cross-validation on the training set. After identifying the optimal number of iterations, the model was retrained on the full training data and its final performance was evaluated on the independent test set. SHAP results indicated that Type of Accident, Accident Object, and Work Process were primary drivers of fatal outcomes, whereas Safety Management Plan and Public/Private Ownership helped lessen severity. Project Cost, Progress Rate, and Number of Workers moderated prediction strength through interactions with key variables. This study clarifies structural relationships among factors affecting accident outcomes using a LightGBM–SHAP framework that captures nonlinear interactions, supporting explainable artificial intelligence (AI)–based safety management and risk monitoring.