AI-estimated heart age aligns with electrical patterns
An observational analysis of 1,206,874 heart recordings from 374,223 patients linked AI-estimated aging categories to distinct patterns in the heart’s electrical signals.
Our read
Biomarkers- Studied:
- Cross-sectional study of 374,223 patients, against normal agers
- What changed:
- Accelerated agers had longer electrical conduction times than normal agers
- The catch:
- Aging categories reflected model predictions rather than an independent measure of cardiovascular aging
- Why it’s aging research:
- It builds or tests a tool for aging research.
Editor of record: Boris Djordjevic · AgingResearch.News
The 374,223 patients in an observational analysis had an average age of 61.7 years; 47.3% were female. Their 1,206,874 electrocardiograms, or ECGs, recorded the heart’s electrical activity. The AI model classified recordings as accelerated or resilient aging when predicted age was more than 9.35 years older or younger than actual age. Other recordings were classified as normal aging.
Among recordings with a rhythm set by the heart’s natural pacemaker, accelerated aging was associated with longer electrical conduction times after accounting for age and sex. Resilient-aging recordings showed the opposite. Across all measured electrical features, accelerated-aging recordings more closely resembled normal-aging recordings matched by AI-predicted age than those matched by actual age.
Why it matters
The authors suggest these electrical signatures may help identify markers of cardiovascular aging that doctors can interpret.
Caveats
The aging categories reflected model predictions rather than an independent measure of cardiovascular aging, and patient differences beyond age and sex could contribute to the patterns. The findings were reported in a conference abstract.
The paper
Yale Sch of Medicine
AHA Scientific Sessions 2026 Abstract, Nov 2, 2026
Presented at AHA Scientific Sessions 2026, abstract 16130
Comments
Be the first to comment.