A recent study presented at Heart Failure 2025, a congress of the European Society of Cardiology (ESC), highlights the promising performance of an artificial intelligence (AI)-enabled electrocardiogram (ECG) algorithm in detecting early signs of heart failure in Kenya. The study specifically focused on identifying left ventricular systolic dysfunction (LVSD), a precursor to heart failure, in a resource-limited setting where access to echocardiography—the gold standard diagnostic tool—is limited.
Led by Dr. Ambarish Pandey from the University of Texas Southwestern Medical Center, this prospective, multicentre screening study assessed nearly 6,000 adult patients across eight healthcare facilities in Kenya. The study employed a validated AI-based ECG algorithm (AiTiALVSD; Medical AI Co, Seoul, South Korea) to detect LVSD by analyzing standard 12-lead ECGs. A threshold score of >0.097 was used to flag individuals at high risk of LVSD.
Of the 5,992 participants, with an average age of 55 and a predominantly female demographic (66%), 65% were classified as high cardiovascular risk. The AI-ECG algorithm detected LVSD in 18.3% of the total participants. Prevalence rates were notably higher among individuals with high Framingham Risk Scores (22.9%) or established cardiovascular disease (32.0%), compared to 9.9% in those with lower risk scores.
A subgroup of 1,444 participants underwent both AI-ECG and echocardiographic assessment. In this comparison, the AI algorithm demonstrated a sensitivity of 95.6%, specificity of 79.4%, and a remarkable negative predictive value of 99.1%, aligning closely with echocardiography results.
According to Dr. Bernard Samia, President of the Kenya Cardiac Society and senior author of the study, these findings suggest that AI-driven ECG tools could serve as scalable, cost-effective screening solutions in underserved areas. Dr. Pandey emphasized the significance of detecting LVSD in nearly 1 in 5 individuals, underlining the hidden burden of heart disease.
The researchers now aim to expand their work through broader screening initiatives across Africa and to explore whether early identification via AI-ECG leads to improved treatment uptake and clinical outcomes.
