AHA Scientific Sessions 2026 AbstractNov 7, 2026, 01:39 GMT

Heart tracing model predicted irregular heartbeat risk across groups

The model scored 0.810 on a measure of distinguishing people who developed atrial fibrillation within 10 years from those who did not.

Paper by Meghana Kamineni, Sam F Friedman, Mahnaz Maddah, +3Shaan Khurshid · Massachusetts General Hosp

Heart cross-sections show jagged electrical activity in a large dark blue-green heart and even waves in a smaller gray heart, surrounded by upright figures.

Our read

Biomarkers

Studied:: Cohort study of 76,967 patients receiving primary or cardiology care, 10 years
What changed:: The model scored AUROC 0.810 for predicting irregular heartbeat, similarly across sexes and racial groups but better in younger patients
The catch:: Race comparisons grouped all non-White participants together, obscuring potential differences between racial groups
In the record:: One of 5 papers on ECG-AI on record: 1 with results in people.
Why it’s aging research:: It bears on staying healthy in midlife and later life.

Editor of record: Boris Djordjevic · AgingResearch.News

Paper

Among 76,967 patients receiving ongoing primary or cardiology care at Massachusetts General Hospital and Brigham and Women’s Hospital, researchers evaluated a heart tracing model in an observational study. Participants averaged 55.2 years of age, and 56.7% were women.

The artificial intelligence model used electrocardiograms, recordings of the heart’s electrical activity, to estimate future atrial fibrillation risk. Atrial fibrillation is an irregular heartbeat. The model’s ability to distinguish people who developed the condition within 10 years from those who did not was similar across sexes and racial groups, and modestly higher in younger patients. In 58,413 participants with data for comparison, its performance across sex, age and race was more consistent than an established clinical risk score based on 11 factors.

Why it matters

Predicting atrial fibrillation across age groups may help identify whom to prioritize for screening and preventive efforts.

Caveats

The analysis tested predictions, not whether model-guided screening improved health. Results came from two hospitals, and race comparisons grouped all non-White participants together.