Development and Validation of a Machine Learning–Based Model for Predicting Textbook Outcome after Minimally Invasive Pancreaticoduodenectomy

Background: Textbook Outcome (TO) reflects overall surgical quality. With the expanding use of minimally invasive pancreaticoduodenectomy (MIPD), reliable prediction of TO is essential. This study aimed to identify predictors of TO after MIPD and develop a machine learning (ML) model. Methods: We retrospectively analyzed 411 patients undergoing MIPD (2017–2023). The Least Absolute Shrinkage and Selection Operator regression was used for feature selection, and ten ML algorithms were compared. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC). Model interpretability was assessed using SHapley Additive exPlanations (SHAP). Results: Of 411 patients, 263 (63.99%) achieved TO. Eight variables were identified as predictive features. Among the ten algorithms, the Random Forest model demonstrated the best discrimination (AUC = 0.86). Conclusions: The Random Forest model accurately predicted TO after MIPD and may assist in individualized preoperative risk stratification and perioperative management.