A Hybrid Machine Learning Model for Predicting Surgical Procedure Duration: Integrating Random Forest and K-Means Clustering

Abstract

Efficient operating room (OR) management depends on the accurate prediction of surgical procedure durations to improve scheduling, enhance patient outcomes, and reduce operational costs. This study presents a hybrid machine learning model that combines Random Forest and K-Means clustering to predict the duration of cholecystectomy procedures. The model is trained using real-world data from the digestive surgery department at Mahmoud El Matri Hospital in Tunis, Tunisia, incorporating patient demographics, surgeon experience, and other contextual factors. Synthetic data generation was also applied to reinforce model reliability. The proposed approach achieved strong performance, with a root mean square error (RMSE) of 0.45 minutes, a mean absolute error (MAE) of 0.36 minutes, and a coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) of 0.99. Comparative analysis with individual models such as Random Forest, K-Means, decision trees, and linear regression confirms the hybrid model’s superior predictive capability. These results demonstrate the potential of the proposed hybrid model as a practical tool for optimizing OR scheduling and improving healthcare resource management.