AI-ECG Prediction of Biventricular Dysfunction

This study developed and externally validated an AI-enhanced ECG model to predict CMR-defined biventricular dysfunction and dilation in patients with congenital heart disease. Using over 9,000 paired ECG–CMR records, the model demonstrated strong and consistent performance across internal and external cohorts, with AUROC values around 0.80–0.89 for detecting ventricular dysfunction and dilation. The findings suggest AI-ECG can help identify high-risk CHD patients and guide timing of advanced imaging, potentially improving clinical management.