A Multimodal and Explainable Machine Learning Approach to Diagnosing Multi-Class Ejection Fraction from Electrocardiograms

Researchers at Hartford HealthCare published an XGBoost-based multimodal framework on arXiv that classifies left ventricular ejection fraction into four clinical strata using combined 12-lead ECG time-series features and structured EHR variables. Trained on 36,784 ECG-echocardiogram pairs, the model achieved one-vs-rest AUROCs of 0.95 for severe, 0.92 for moderate, 0.82 for mild, and 0.91 for normal LVEF, outperforming ECG-only and EHR-only baselines. SHAP attributions provide feature-level explainability, critical for clinical trust. Primary care and resource-limited settings stand to gain most, as this approach bypasses echocardiography dependency entirely.