Prediction of Surgery Duration Based on Machine Learning Algorithms and Its Practical Application in a General Hospital in China

This study aims to optimise operating room scheduling and improve hospital operational efficiency by predicting surgery durations using machine learning algorithms. Traditional methods often face challenges in accuracy, while machine learning models offer superior predictive performance. Using real‐world operating room data from a large general hospital in China, the study compares various machine learning algorithms and selects the XGBoost model as the most effective predictive framework. Three types of models were developed: an all‐inclusive model, a department‐specific model, and a doctor‐specific model. The department‐specific model demonstrated the highest accuracy, outperforming the others. The results were applied to the hospital's surgical centre scheduling process, significantly enhancing operating room resource utilisation. The study highlights the importance of data preprocessing and feature selection in improving prediction accuracy. Overall, machine learning‐based surgery duration prediction can effectively address challenges in surgical scheduling and provide data‐driven support for hospital operational management.