AI-Enabled Digital Auscultation for Detecting Heart Failure with Reduced Ejection Fraction in Sub-Saharan Africa: The DAMSUN-HF Study
This prospective diagnostic–implementation study evaluated whether an AI-enabled, Bluetooth-connected digital stethoscope platform (Eko SENSORA™ with the Low EF algorithm) can detect heart failure with reduced ejection fraction (HFrEF) in low-resource settings. Adults presenting with dyspnea, orthopnea, or edema at four Ghanaian hospitals underwent point-of-care auscultation; AI results (LVEF ≤40% vs >40%) were confirmed by blinded transthoracic echocardiography within 7 days, demonstrating real-world feasibility in a hub-and-spoke workflow. The system correctly identified 62/64 true low-EF cases with 97% sensitivity (95% CI 89–99), 76% specificity (63–86), AUC 0.92, PPV 85%, and NPV 94%, with 93% overall workflow adherence (TTE in 97%, 95% within 7 days; >90% cardiologist review within 48 hours). Findings support AI-enabled auscultation as a scalable, low-cost adjunct to imaging that can accelerate recognition of systolic dysfunction and streamline referral in settings with limited echocardiography capacity.
