Machine-learning model predicts postoperative infection risk after surgery

A machine-learning model, specifically the extreme gradient boost (XGBoost), was developed to predict postoperative infection following cytoreductive surgery and hyperthermic intraperitoneal chemotherapy with splenectomy. The model demonstrated excellent prediction accuracy, achieving an area under the curve (AUC) of 0.910 in preliminary testing and 0.823 in external validation. Data was analyzed from a cohort of 1,016 patients, with 21% experiencing postoperative infections, highlighting the algorithm’s potential for early diagnosis and treatment initiation.