Heart tracing model identifies patients at higher death risk
Five-year mortality was 29.8% in the model’s highest filling-pressure category versus 14.2% in its lowest among patients with grade 1 heart relaxation abnormalities.

Our read
Biomarkers- Studied:
- Cohort study of 18,243 patients with grade 1 diastolic dysfunction, median 4.9 years, against Group A (Dose-response)
- What changed:
- 5-year mortality was 29.8% in Group C against 14.2% in Group A
- The catch:
- Higher-risk patients were older and had more heart failure and chronic kidney disease
Editor of record: Boris Djordjevic · AgingResearch.News
At Mayo Clinic, 18,243 patients with grade 1 diastolic dysfunction—impaired heart relaxation identified by ultrasound—were assessed in an observational cohort study. A previously published artificial intelligence model analyzed electrocardiograms, recordings of the heart’s electrical activity, and assigned patients to three estimated heart filling-pressure categories.
Median follow-up was 4.9 years. Five-year mortality was 14.2% in the lowest category, 18.7% in the middle category and 29.8% in the highest. The association between the highest category and greater mortality persisted after adjustment for age, sex and other illnesses. Mortality risk also rose continuously with the model’s filling-pressure probability. The authors suggested that the model may complement ultrasound for assessing risk in patients with mild abnormalities.
Why it matters
For older people with heart relaxation abnormalities, the work addresses whether electrical recordings can add information about mortality risk beyond an ultrasound grade.
Caveats
Age and illness differed across groups, so unmeasured differences in patients’ health may still have contributed to the mortality gap. This cohort, reported in a conference abstract, did not test whether model-guided care improved outcomes.
- AI-ECG filling-pressure category
- Grade 1 diastolic dysfunction
- AI-ECG filling-pressure probability
- E/E′ ratio
- Left atrial volume index
- Humans
The paper
Mayo Clinic
AHA Scientific Sessions 2026 Abstract, Nov 2, 2026
Presented at AHA Scientific Sessions 2026, abstract 8926




Comments
Be the first to comment.