AI-Enabled ECG Algorithm Enables Early Detection of Heart Failure

Early detection of left ventricular systolic dysfunction (LVSD) is a key strategy for identifying patients who are at risk of developing heart failure. However, access to echocardiography, the gold standard for diagnosing LVSD, is limited in resource-poor settings. New research presented at Heart Failure 2025 highlights how an artificial intelligence (AI)-based electrocardiogram (ECG) algorithm has shown promise in detecting heart failure early among individuals seeking healthcare in Kenya.

The study, led by researchers from the University of Texas Southwestern Medical Center (Dallas, TX, USA), explored whether LVSD could be evaluated using an AI-powered ECG algorithm as a scalable solution for large-scale population screening. The prospective, cross-sectional, multicenter study involved adult patients who visited eight healthcare facilities across Kenya. The study assessed cardiovascular risk based on previous cardiovascular disease (CVD) history or a Framingham Risk Score (FRS) greater than 10%. All participants underwent a 12-lead ECG, and LVSD (<40%) was evaluated using AiTiALVSD, an AI-ECG algorithm developed by Medical AI Co (Seoul, Republic of Korea). The algorithm determined the probability of LVSD using a predefined risk threshold of >0.097. A subset of participants also underwent LVSD assessments using both the AI-ECG algorithm and echocardiography for comparison.

The study involved 5,992 participants with a mean age of 55 years, two-thirds of whom were women (66%) and 65% were categorized as being at high cardiovascular risk. The AI-ECG algorithm identified a prevalence of LVSD in 18.3% of participants, with higher rates found among those with high Framingham Risk Scores (22.9%) or pre-existing cardiovascular disease (32.0%), compared to those with low FRS (9.9%). Among 1,444 participants with both AI-ECG and echocardiography assessments, LVSD was confirmed in 14.1% of cases via echocardiography. The AI-ECG algorithm performed exceptionally well, with sensitivity of 95.6%, specificity of 79.4%, and a negative predictive value of 99.1%, demonstrating its reliability when compared to echocardiography.

“It was striking that the AI-ECG algorithm identified LVSD in almost 1 in 5 individuals, highlighting the large population at risk of heart failure,” said study presenter Dr. Ambarish Pandey from UT Southwestern Medical Center. “Given that the AI-ECG algorithm performed well against the gold standard method, we would now like to conduct larger screening studies across several countries in Africa. It will also be important to investigate whether identification of LVSD leads to greater use of evidence-based therapies.”

Related Links:
UT Southwestern Medical Center
Medical AI Co