Deep Learning-Based Prediction Model of Acute Kidney Injury Following Coronary Artery Bypass Grafting in Coronary Heart Disease Patients: A Multicenter Clinical Study from China

Biao Hou¹Tingting Liu¹Pengyun Yan²Yuqing Wang³Xuejian Hou¹Liang Li⁴Haiping Yang⁵Lin Chen⁵Taoshuai Liu¹Zhang Kui¹Shijun Xu¹Yang Li^1*Ran Dong^1*

• ¹Beijing Anzhen Hospital, Capital Medical University, Chaoyang District, China
• ²The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, Shaanxi, China
• ³Jinan No. 3 People's Hospital, Jinan, Shandong Province, China
• ⁴Handan First Hospital, Handan, Hebei Province, China
• ⁵Beijing Luhe Hospital, Capital Medical University, Beijing, Beijing Municipality, China

Off-pump coronary artery bypass grafting (OPCABG) is an alternative to traditional CABG, avoiding cardiopulmonary bypass. However, acute kidney injury (AKI) remains a common complication, with incidence rates ranging from 5% to 42%, significantly affecting postoperative outcomes.This study aimed to develop a robust risk prediction model for post-OPCABG AKI using machine learning (ML) techniques. We conducted a multicenter retrospective study with 3,043 coronary artery disease (CAD) patients, with an overall AKI incidence of 15.28%. The cohort was divided into a training set (n=2,130) and a validation set (n=913), and an external validation cohort of 878 patients was included. Five ML methods -SVM, Decision Tree (DT), Random Forest (RF), AdaBoost, and XGBoostwere employed to predict AKI risk. The XGBoost model showed the highest performance, with an area under the curve (AUC) of 0.88, sensitivity of 82%, and specificity of 83% in the internal validation set. In the external validation set, it achieved an AUC of 0.84, sensitivity of 74%, and specificity of 90%. The model utilized 26 predictive features, including patient demographics and preoperative laboratory values. The XGBoost model outperformed other methods (SVM, DT, RF, and AdaBoost) in both internal and external validation, demonstrating its robustness and generalizability. By integrating diverse patient data from multiple institutions, our model improved AKI risk assessment and identified novel predictive factors using advanced ML algorithms. These findings highlight the potential of machine learning models in enhancing AKI risk prediction and supporting personalized management strategies to improve outcomes in OPCABG patients.