Ambulatory Surgery Ensemble: Predicting Adult and Pediatric Same-Day Surgery Cases Across Specialties.

Objective: To develop an ensemble model using case-posting data to predict which patients could be discharged on the day of surgery. Background: Few models have predicted which surgeries are appropriate for day cases. Increasing the ratio of ambulatory surgeries can decrease costs and inpatient bed utilization while improving resource utilization. Methods: Adult and pediatric patients undergoing elective surgery with any surgical specialty in a multisite academic health system from January 2021 to December 2023 were included in this retrospective study. We used surgical case data available at the time of case posting and created 3 gradient-boosting decision tree classification models to predict case length (CL) less than 6 hours, postoperative length of stay (LOS) less than 6 hours, and home discharge disposition (DD). The models were used to develop an ambulatory surgery ensemble (ASE) model to predict same-day surgery (SDS) cases. Results: The ASE achieved an area under the receiver operating characteristic curve of 0.95 and an average precision of 0.96. In total, 139,593 cases were included, 48,464 of which were in 2023 and were used for model validation. These methods identified that up to 20% of inpatient cases could be moved to SDS and identified which specialties, procedures, and surgeons had the most opportunity to transition cases. Conclusions: An ensemble model can predict CL, LOS, and DD for elective cases across multiple services and locations at the time of case posting. While limited in its inclusion of patient factors, this model can systematically facilitate clinical operations such as strategic planning, surgical block time, and case scheduling.