A Data-Driven Approach for Predicting Remaining Intra-Surgical Time and Enhancing Operating Room Efficiency

Purpose: Efficient scheduling in Operating Rooms (ORs) is essential for optimizing corresponding costs and enhancing customer satisfaction in healthcare systems. Design/methodology/approach: Conventional static scheduling methods rely on fixed historical surgery times and often lead to inefficient resource utilization and increased costs due to inaccurate predictions of surgical durations. In this regard, this paper introduces an innovative method that employs Convolutional Neural Networks (CNNs) to predict the remaining intra-surgical time through binary classification for the Gallbladder Dissection phase and to dynamically manage OR schedules. The study, although focused on laparoscopic cholecystectomy procedures, demonstrates a method adaptable to other laparoscopic surgeries. The dataset comprises labeled laparoscopic cholecystectomy videos (time labels for different phases) used to train and evaluate the CNN. Findings: Results show that the proposed method reduces patient waiting times by an average of 87.3% and eliminates OR idle time compared to traditional fixed-time scheduling methods. Originality/value: This paper introduces a new data-driven approach for predicting remaining intrasurgical time and enhancing OR efficiency. The study’s novelty lies in its use of Convolutional Neural Networks (CNNs) to predict surgery completion times, a method that has not been extensively explored in this context. By providing accurate forecasts, this approach allows nurses to prepare for the next patient more efficiently and enables dynamic rescheduling when surgeries deviate from their planned timelines. This combination contributes to improved OR utilization and enhances patient satisfaction, offering a practical and innovative solution to a common challenge in surgical workflow management.