Machine Learning-Based Discrimination of Cardiovascular Outcomes in Patients With Hypertrophic Cardiomyopathy

Area of Science:

• Cardiology
• Biomedical Engineering
• Data Science

Background:

• Traditional risk stratification for hypertrophic cardiomyopathy (HCM) has limitations.
• Novel approaches are needed to improve patient outcomes.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting major adverse cardiovascular events (MACE) in HCM patients.
• To identify key clinical and echocardiographic features influencing cardiovascular risk in HCM.

Main Methods:

• Utilized data from 2,111 HCM patients, analyzing 25 clinical and echocardiographic features.
• Developed four ML models, selecting the best performing logistic regression model via cross-validation (AUROC).
• Employed SHapley Additive exPlanations (SHAP) to determine feature importance for MACE, all-cause death, heart failure admission, and stroke.

Main Results:

• The logistic regression model demonstrated strong predictive performance for MACE (AUROC=0.800) and its components.
• External validation confirmed the model's discriminant ability for MACE (AUROC=0.768), all-cause death (AUROC=0.750), and HF-adm (AUROC=0.806).
• SHAP analysis identified left atrial diameter and hypertension as significant predictors across all adverse outcomes.

Conclusions:

• ML models effectively discriminate cardiovascular events in HCM patients.
• The models provide valuable insights into key risk factors, enhancing current risk stratification strategies.
• Left atrial diameter and hypertension are crucial variables for predicting adverse outcomes in HCM.