An Explainable Machine Learning Approach Reveals Prognostic Significance of Right Ventricular Dysfunction in Nonischemic Cardiomyopathy

Abstract

Risk stratification in patients with idiopathic nonischemic cardiomyopathy (NICM) remains challenging caused by heterogeneous clinical presentations and unpredictive disease progression. Patients with NICM are prone to frequent hospitalization caused by worsening heart failure (HF) symptoms. Indeed, frequent HF hospitalization remains a major health care burden and hospital readmissions due to HF is being used as a value-based metric by the Center for Medicare and Medicaid Services. Recent studies have shown that left ventricular ejection fraction (LVEF) alone, a popular major marker of adverse outcomes, is less sensitive at identifying those in need of frequent hospitalization than predicting arrhythmia. Myocardial fibrosis, imaged using late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR), provides both diagnostic and prognostic information in patients with NICM. Several studies reported both presence and extent of myocardial fibrosis as significant indicators of HF hospitalization. Gulati et al recently demonstrated that right ventricular (RV) systolic dysfunction in patients with dilated cardiomyopathy independently predicts transplant-free survival and adverse HF outcomes.

Salah Assana

Research Assistant ll

I am a Artificial Intelligence and Healthcare enthusiast with a background in machine learning, signal processing and medical imaging.