Predicting operative time for metabolic and bariatric surgery using machine learning models: a retrospective...

Area of Science:

• Surgical Informatics
• Machine Learning in Healthcare
• Bariatric Surgery Outcomes

Background:

• Accurate prediction of operative time is crucial for efficient operating room management and surgical scheduling.
• Metabolic and bariatric surgery (MBS) procedures have variable operative times, necessitating predictive tools.
• Existing methods for predicting MBS operative time may lack precision.

Purpose of the Study:

• To develop and compare machine learning (ML) models for predicting operative time in metabolic and bariatric surgery (MBS).
• To identify key factors influencing operative time in MBS.
• To evaluate the performance of various ML algorithms for this predictive task.

Main Methods:

• Utilized the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program database (2016-2020).
• Developed and compared multiple ML models: linear regression, random forest, support vector machine, gradient-boosted tree, and XGBoost.
• Evaluated model performance using mean absolute error, root mean square error, and R-squared score; identified important variables via Shapley Additive exPlanations.

Main Results:

• The XGBoost model exhibited superior performance in predicting operative time, evidenced by the lowest root mean square error and highest R-squared score.
• Sleeve gastrectomy and laparoscopic approaches were associated with shorter operative times compared to Roux-en-Y gastric bypass and robotic-assisted approaches, respectively.
• Surgery type and surgical approach emerged as the most significant predictors of operative time.

Conclusions:

• The XGBoost model provides the most accurate prediction of operative time for metabolic and bariatric surgery among the evaluated ML models.
• These predictive capabilities can enhance operating room scheduling and inform the development of clinical software tools for MBS.
• The findings highlight the potential of ML to optimize surgical workflow and resource allocation in bariatric surgery.