New Echocardiographic Risk Score for HCM Patients Follow-up

Current risk stratification in hypertrophic cardiomyopathy (HCM) often overlooks the full diagnostic potential of echocardiography and provides limited support for longitudinal patient monitoring. This study proposes a novel risk score based only on 26 conventional echocardiographic parameters, trained using machine learning to predict 5-year composite cardiovascular events. We retrospectively analyzed 1,061 HCM patients from the SHARE registry, applying logistic regression, support vector machine, random forest, and gradient boosting classifiers. The composite endpoint included heart failure progression and arrhythmic events. Logistic regression achieved the best performance, with a balanced accuracy of 73.6%, sensitivity of 72.2%, and specificity of 75.0% using nested 5-fold cross-validation. Moreover, longitudinal analysis in patients with serial echocardiographic follow-ups showed that the predicted risk score increased progressively in 83.7% of those who later experienced an event, suggesting value in dynamic risk tracking. These findings highlight the potential of echocardiography-driven machine learning tools to enhance individualized HCM management through both static risk assessment and dynamic followup.