A machine learning model predicts anastomotic leakage after surgery

An advanced interpretable machine learning model, based on the XGBoost algorithm, shows promising results in predicting anastomotic leakage (AL) following rectal cancer resection. Utilizing data from multiple centers, the model achieved an AUC of 0.984 in the test set, with high accuracy, sensitivity, and specificity. Serum calcium ion levels emerged as a critical predictor. This framework aids clinicians in optimizing perioperative strategies and enhances decision-making in surgical contexts.