Machine Learning Based Prediction of Postoperative Acute Kidney Injury Risk in Coronary Artery Bypass Grafting Patients.

Background: Coronary artery bypass grafting (CABG) is key for severe coronary artery disease, but postoperative acute kidney injury (AKI) may increase mortality and prolong hospital stays. Reliable models for early prediction of post-CABG AKI remain lacking. Methods: Data of 520 CABG patients (September 2021-December 2024) from the Affiliated Hospital of Xuzhou Medical University were collected, and the patients were divided into a training group (70%, for model building) and a validation group (30%). Key variables were screened through Least Absolute Shrinkage and Selection Operator (LASSO) regression, followed by the construction of six machine learning models: Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Logistic Regression (LR), Light Gradient Boosting Machine (LightGBM), Softmax Regression, and Support Vector Machine (SVM). The SHapley Additive exPlanations (SHAP) was used to quantify feature importance. Results: The incidence of post-CABG AKI was 25.96%, and the median age of patients in the AKI group was significantly higher than that in the non-AKI group (66.09 ± 8.15 vs 64.32 ± 7.76, p = 0.025). In the training group, the XGBoost model using the top 5 important variables outperformed other models (Area Under the Curve [AUC] = 0.89, 95% Confidence Interval [CI]: 0.86-0.91), followed by the LightGBM model using the top 5 important variables and the RF model using the top 5 important variables (both had an AUC of 0.88; 95% CI: 0.85-0.90 and 0.85-0.91, respectively). In the validation group, the LR model using the top 15 important variables and the Softmax Regression model using the top 15 important variables maintained the highest stability (both had an AUC of 0.86, 95% CI: 0.79-0.92). SHAP analysis confirmed that estimated glomerular filtration rate (eGFR), intraoperative epinephrine use and calcium levels were the top three predictive factors. Conclusion: The machine learning models constructed in this study can effectively predict post-CABG AKI, facilitating early identification of high-risk patients.