| Abstract:Objective: This study aims to predict the risk of postoperative acute kidney injury (AKI) after gastric resection surgery using machine learning methods, including decision trees, logistic regression, and random forests. Additionally, it seeks to conduct an in-depth analysis of key predictive factors.
Methods: Patients who underwent general anesthesia for gastric resection surgery at Zhejiang Provincial People's Hospital between January 2021 and December 2022 were selected. Demographic data, intraoperative, preoperative, and postoperative laboratory examination information were collected. Three machine learning models—decision tree, logistic regression, and random forest—were employed. Through feature selection and model performance evaluation, the optimal predictive model was determined.
Conclusion: The random forest model exhibited the best performance in terms of AUC (AUC=0.91, 95% CI: (0.8074, 0.9686)), demonstrating its adaptability to different data and complex relationships. Important predictive factors included male gender, diabetes, preoperative neutrophil classification, intraoperative bicarbonate, crystalloid fluid intake, and postoperative high-sensitivity C-reactive protein. The study provides new possibilities for personalized management and offers guidance for related research and clinical practice. |
