Journal of anesthesia and translational medicine · 2026
Prediction of postoperative acute kidney injury in patients undergoing off-pump coronary artery bypass grafting: A machine learning model.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Wang Sen, Wang Wenli, Bu Yuchen, Liu Hanyu, Yang Chun, Wang Yuanyuan
Аннотация
BACKGROUND: The incidence of cardiac surgery-associated acute kidney injury (CSA-AKI) is 26.0%-28.5%. Among cardiac procedures, off-pump coronary artery bypass grafting (OPCABG) is a major contributor to CSA-AKI. Early detection and prompt intervention for high-risk patients are crucial. We therefore developed a machine learning model to predict OPCABG-associated acute kidney injury (AKI), aiming to inform perioperative clinical decision-making.
METHODS: This retrospective study analyzed 938 patients who underwent OPCABG from June 2023 to January 2025. We preprocessed baseline and intraoperative time-series data separately. Using Tsfresh in Python, we extracted time-series intraoperative features, which were then integrated with selected features for model retraining. The final OPCABG-AKI risk prediction model was established by selecting the optimal model based on accuracy and AUC metrics. SHAP analysis was employed to rank feature importance and identify key risk factors.
RESULTS: Of the 938 patients, 210 (22.39%) developed OPCABG-AKI. The XGBoost model outperformed others in predicting OPCABG-AKI, achieving an AUC of 0.982, accuracy of 0.916, and precision of 0.983 in external validation. SHAP analysis identified Cystatin C (Cys-C) as the top predictor, with NT-proBNP levels ranking second.
CONCLUSIONS: The XGBoost model showed outstanding performance in predicting OPCABG-AKI. Selecting features from preoperative data and integrating them with intraoperative data significantly improved prediction accuracy. This model enables clinicians to stratify patient risks and make informed decisions regarding OPCABG-AKI, promoting early patient recovery.
METHODS: This retrospective study analyzed 938 patients who underwent OPCABG from June 2023 to January 2025. We preprocessed baseline and intraoperative time-series data separately. Using Tsfresh in Python, we extracted time-series intraoperative features, which were then integrated with selected features for model retraining. The final OPCABG-AKI risk prediction model was established by selecting the optimal model based on accuracy and AUC metrics. SHAP analysis was employed to rank feature importance and identify key risk factors.
RESULTS: Of the 938 patients, 210 (22.39%) developed OPCABG-AKI. The XGBoost model outperformed others in predicting OPCABG-AKI, achieving an AUC of 0.982, accuracy of 0.916, and precision of 0.983 in external validation. SHAP analysis identified Cystatin C (Cys-C) as the top predictor, with NT-proBNP levels ranking second.
CONCLUSIONS: The XGBoost model showed outstanding performance in predicting OPCABG-AKI. Selecting features from preoperative data and integrating them with intraoperative data significantly improved prediction accuracy. This model enables clinicians to stratify patient risks and make informed decisions regarding OPCABG-AKI, promoting early patient recovery.