Deep Survival Modeling With Echocardiographic Myocardial Texture Radiomics for Prediction of Major Adverse Cardiovascular Events in Hypertrophic Cardiomyopathy.

Hypertrophic cardiomyopathy (HCM) carries substantial risk of major adverse cardiovascular events (MACE), yet current tools do not fully exploit quantitative myocardial texture from routine echocardiography. We retrospectively studied 413 patients with HCM from two tertiary centers to develop and externally validate a deep survival model integrating transthoracic echocardiographic texture radiomics with clinical and conventional echocardiographic variables. Patients from one center (n = 308) were used for development and internal testing and those from the other (n = 105) for external validation. The best-performing model was selected by cross-validated Harrell C-index and interpreted using Shapley additive explanations. During median follow-up of 30.7 and 35.6 months, 83 patients (20.1%) experienced a first MACE. The Combined model showed the highest concordance and net clinical benefit, with age, RadScore, left atrial diameter, and maximal wall thickness as leading contributors. This model may provide an accessible approach to individualised MACE risk stratification in HCM.