A significant study utilizing machine learning has identified heart failure with preserved ejection fraction (HFpEF) as a major independent predictor of adverse outcomes in patients with hypertrophic cardiomyopathy (HCM). This highlights a previously underestimated high-risk patient subgroup.
HCM, a genetic condition causing abnormal thickening of the heart muscle, has diverse clinical trajectories. HFpEF in HCM is increasingly recognized but poorly understood, necessitating refined risk stratification.
The multicenter retrospective study analyzed 2,802 HCM patients, finding nearly half (47.8%) had HFpEF. Even after balancing baseline characteristics, HFpEF in HCM remained strongly linked to worse event-free survival. Risk stratification using the H₂FPEF score further revealed significantly poorer outcomes for high-risk patients.
Researchers developed four machine learning models to improve prognostication, with a random forest model achieving the highest accuracy. Key predictors identified were HFpEF status and B-type natriuretic peptide (BNP) levels, with a non-linear relationship observed for BNP, indicating potentially underestimated risk at higher concentrations.
These findings suggest HFpEF in HCM warrants closer clinical attention. Integrating scoring systems, biomarkers, and machine learning offers a more nuanced framework for risk assessment. While external validation is needed, data-driven approaches show promise for refining prognostic assessment and guiding personalized management in HCM.