Machine Learning Model for Predicting Postoperative Complications in Pediatric Simple Congenital Heart Disease with Right Vertical Infra-Axillary Incision.

BACKGROUND

This study aimed to develop and validate a machine learning model to predict postoperative complications in pediatric simple congenital heart disease (CHD) patients undergoing right vertical infra-axillary incision (RVIAI).

METHODS

A retrospective dataset of 638 patients who underwent treatment for ventricular septal defect and/or atrial septal defect via RVIAI at our hospital between August 2020 and August 2023 was collected. A total of 35 preoperative and intraoperative variables were used to construct 190 machine learning models. The optimal model was selected based on the highest mean C-index. Independent risk factors identified by the optimal model were ranked according to their importance. Kaplan-Meier analysis was used to compare the incidence of postoperative complications between different risk groups. Model performance was evaluated using the area under the receiver operating characteristic curve (ROC).

RESULTS

The optimal model, which combined Elastic Net (alpha = 0) and Gradient Boosting Machine, identified 18 baseline variables associated with postoperative complications. The top five predictors were defect size, globulin, activated partial thromboplastin time, red blood cell count, and blood urea nitrogen. Kaplan-Meier curves showed that postoperative complication rates were significantly higher in the high-risk group than in the low-risk group (p < 0.0001). The model demonstrated good discrimination, with area under the curve (AUC) values on postoperative days 5, 10, 15, and 20 remaining above 0.78 in both the training and test sets.

CONCLUSIONS

This machine learning model provides a potential predictive tool for assessing postoperative risk in simple CHD patients undergoing RVIAI and may support more targeted perioperative management.