Estimation of Surgery Durations Using Machine Learning Methods-A Cross-Country Multi-Site Collaborative Study

Abstract

The scheduling of operating room (OR) slots requires the accurate prediction of surgery duration. We evaluated the performance of existing Moving Average (MA) based estimates with novel machine learning (ML)-based models of surgery durations across two sites in the US and Singapore. We used the Duke Protected Analytics Computing Environment (PACE) to facilitate data-sharing and big data analytics across the US and Singapore. Data from all colorectal surgery patients between 1 January 2012 and 31 December 2017 in Singapore and, 1 January 2015 to 31 December 2019 in the US were used, and 7585 cases and 3597 single and multiple procedure cases from Singapore and US were included. The ML models were based on categorical gradient boosting (CatBoost) models trained on common data fields shared by both institutions. The procedure codes were based on the Table of Surgical Procedure (TOSP) (Singapore) and the Current Procedural Terminology (CPT) codes (US). The two types of codes were mapped by surgical experts. The CPT codes were then transformed into the relative value unit (RVU). The ML models outperformed the baseline MA models. The MA, scheduled durations and procedure codes were found to have higher loadings as compared to surgeon factors. We further demonstrated the use of the Duke PACE in facilitating data-sharing and big data analytics.

Artificial Intelligence in Operating Room Management

“…Among the 22 studies analyzed [ 8 – 29 ], sixteen primarily focused on predicting the duration of surgical cases [ 8 – 13 , 15 – 24 ], three centered on predicting the length of stay in the PACU [ 25 – 27 ]. One study addressed both aspects [ 14 ], while only two studies examined the identification of surgical cases at high risk of cancellation [ 28 , 29 ].…”

“…For instance, Hassanzadeh et al [ 11 ] predicted daily operating theatre arrivals with 90% accuracy, optimizing staffing and resource allocation. Several studies, including those from Bartek et al [ 8 ] and Lam et al [ 13 ], emphasized the importance of tailoring ML models to individual surgeons or considering additional patient and surgery-related factors.…”

Development, Deployment, and Implementation of a Machine Learning Surgical Case Length Prediction Model and Prospective Evaluation

“…Fifth, our model was developed using the data from a single health system. Although we hypothesize that using similar feature sets and the similarity cascade in other health systems should perform comparably, 19 our model needs to be validated at other institutions. Sixth, the features set in the current model are from case posting data.…”

Machine Learning for the Prediction of Procedural Case Durations Developed Using a Large Multicenter Database: Algorithm Development and Validation Study

“…In short, machine learning aims to extract patterns of knowledge from data, the benefit being the ability to process large volumes of disparate data, exploring potentially nonlinear interactions that may challenge the required assumptions of conventional analysis. Nonetheless, current studies have been limited to single or few institutions, smaller sample sizes (between 400 and 80,000 cases), specific surgical subpopulations (robotic [10], colorectal [11], and pediatric [12]), or the use of proprietary algorithms [10][11][12][13][14][15][16]; for example, the study by Lam et al [11] was multicenter but had approximately 10,000 colorectal cases. The studies by Tuwatananurak et al [13] and Rozario and Rozario [14] used proprietary tools, which may be useful for adoption but do not permit the same level of transparency or explainability as other methods.…”