Machine Learning for Predicting Heart Failure Progression in Hypertrophic Cardiomyopathy

Area of Science:

• Cardiology
• Biomedical Engineering
• Data Science

Background:

• Advanced heart failure (HF) is a common complication in hypertrophic cardiomyopathy (HCM).
• Current methods for identifying HCM patients at risk of HF are limited.
• Early identification of high-risk patients is crucial for timely intervention.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for predicting the risk of advanced HF in HCM patients.
• To identify key clinical and imaging predictors of HF progression in HCM.
• To provide a tool for personalized risk stratification in HCM management.

Main Methods:

• Utilized a cohort of 1,427 HCM patients with baseline clinical and echocardiographic data.
• Employed the least absolute shrinkage and selection operator (LASSO) for variable selection.
• Developed an ensemble ML model, including logistic regression, for risk prediction.
• Validated the model on a separate subset of the patient cohort.

Main Results:

• The ML model identified 17 key predictors of advanced HF, including sex, NYHA class, HCM type, LV wall thickness, LVEF, HF symptoms, comorbidities, and medications.
• The model demonstrated strong predictive performance with a c-statistic of 0.81 in the validation set.
• Achieved 74% accuracy, 80% sensitivity, and 72% specificity in identifying high-risk patients.
• Model performance was consistent across different sex and age demographics.

Conclusions:

• Machine learning analysis of clinical and imaging parameters accurately predicts 5-year risk of progressive HF in HCM patients.
• The developed ML model offers a valuable tool for risk stratification and personalized management of HCM.
• Identifying key predictors can guide future research and clinical strategies for HF prevention in HCM.