1 September 2018 in International Journal of Production Economics
2 July 2026 in SSRN Electronic Journal
Purpose Efficient operating room (OR) scheduling depends on accurate prediction of duration of surgery (DOS) and effective sequencing strategies. In high-volume arthroplasty procedures, reliance on averages or surgeons’ DOS estimates can cause inefficiencies in resource allocation. This study develops an integrated framework combining machine learning (ML) DOS prediction with discrete-event simulation (DES) to support evidence-based sequencing and scheduling decisions. Design/methodology/approach A dataset of 3,233 hip and knee arthroplasty cases was analyzed using preoperative clinical, operational, and temporal features. Multiple ML models were tested for DOS prediction; performance and feature importance were evaluated using accuracy metrics, SHAP values, and partial dependence. Predicted DOS values were then applied within DES to compare sequencing strategies and assess impacts on completion times, and overtime in marginal conditions. Findings Extreme Gradient Boosting achieved the best predictive accuracy (Mean Absolute Error = 12.86, Root Mean Squared Error = 16.67, Buffer Accuracy = 68.73). Surgeon, procedure type, and temporal variables were key predictors. Simulation experiments revealed that no single sequencing strategy consistently minimized finish times; effectiveness depended on patient-specific and contextual factors. Marginal condition analysis identified opportunities for improved OR utilization and reduced overtime through tailored sequencing policies. Originality/value This study uniquely integrates ML-based DOS prediction with simulation-driven sequencing evaluation in arthroplasty surgery. The framework advances beyond subjective or average-based planning by enabling data-driven, case-specific scheduling. The approach is adaptable to other surgical specialties and can incorporate additional predictors or real-time data, offering a novel pathway toward intelligent scheduling systems that balance clinical priorities with operational efficiency.