AI-ECG Predicts Long-Term Risk of Cardiac Events and Mortality
AI analysis of single-lead ECG helps identify high-risk patients — even when heart function appears normal.
Title: Artificial Intelligence Analysis of the Single-Lead ECG Predicts Long-Term Clinical Outcomes
Authors: Abdullah Alrumayh, Patrik Bächtiger, Arunashis Sau, et al.
Journal: European Heart Journal – Digital Health, 2025
This study evaluated whether Low EF AI, an artificial intelligence (AI) tool originally developed to detect reduced ejection fraction (EF ≤ 40%) from a single-lead electrocardiogram (ECG), could also predict long-term cardiovascular risk. Researchers followed 1,007 patients who received routine heart ultrasounds and simultaneous ECGs analyzed by Low EF AI. About one-third (33.7%) were flagged by the AI as having possible low EF.
After two years:
- Patients flagged by Low EF AI had a 34.2% rate of major adverse cardiovascular events (MACE) compared to 11.9% in those not flagged
- Mortality was 23.0% in the AI-positive group, versus 9.6% in the AI-negative group
Even among patients with preserved EF (≥ 50%) on ultrasound, those flagged by Low EF AI had worse outcomes:
- MACE: 27.2% vs. 11.9%
- Mortality: 20.4% vs. 9.6%
These findings suggest that Low EF AI may identify high-risk patients even when heart function appears normal on imaging. The tool could support early, point-of-care risk stratification, pending further validation.
