AI heart-tracing screens linked to later heart failure
Positive combined screens were linked to 3.4-fold and 4.3-fold relative heart failure risks versus negative screens at Massachusetts General and Emory, respectively.
Our read
Biomarkers- Studied:
- Cohort study of patients without prior heart failure, 3.6 and 5.9 years, against negative screens
- What changed:
- Heart failure risk was higher with positive combined screens: 3.4-fold at Massachusetts General and 4.3-fold at Emory versus negative screens
- The catch:
- Health differences beyond the factors included in the analysis may have contributed to the links
- Why it’s aging research:
- It bears on staying healthy in midlife and later life.
Editor of record: Boris Djordjevic · AgingResearch.News
Patients without a previous heart failure diagnosis—178,537 at Massachusetts General Hospital and 55,010 at Emory University Hospital—were assessed in an observational study of electrocardiograms, recordings of the heart’s electrical activity. Researchers applied seven existing artificial intelligence models designed to detect structural heart disease, which affects the heart’s muscle or valves. They used the models without retraining or adjusting them for either hospital.
Over median follow-ups of 3.6 and 5.9 years, respectively, 5.4% and 18.0% developed heart failure. Positive results from all seven models were linked to later heart failure at both hospitals. The combined model’s links remained after accounting for age, sex, high blood pressure, diabetes and obesity. The disease-specific models also improved risk prediction beyond those factors.
Why it matters
Heart failure is an age-related disease. The findings suggest that models built to detect existing heart problems may also offer a way to assess future heart failure risk.
Caveats
Health differences beyond the factors included in the analysis may have contributed to the links. These conference-abstract findings did not establish whether acting on screening results would prevent heart failure.
The paper
Yale Sch of Medicine
AHA Scientific Sessions 2026 Abstract, Nov 2, 2026
Presented at AHA Scientific Sessions 2026, abstract 11917
Comments
Be the first to comment.