Heart failure & cardiomyopathies
Predicting Heart Failure From 12-Lead ECGs Using AI: A HeartShare/AMP-HF Pooled Cohort Analysis
Authors' conclusion
The addition of ECG-AI to PREVENT-HF improved discrimination of near-term HF risk.
Study context
- Population
- Participants of the Framingham Heart Study, Multi-Ethnic Study of Atherosclerosis and Cardiovascular Health Study with data for both ECG-AI and PREVENT-HF
- Participants
- 14,126 participants
- Exposure
- ECG-AI added to PREVENT-HF vs PREVENT-HF alone
- Outcome
- Incident HF prediction (Harrell's C-statistic and net reclassification improvement)
- Follow-up
- 1, 3, 5 and 10 years
Methodological detailspooled data from 3 cohorts (Framingham Heart Study, MESA, Cardiovascular Health Study)
- Pooled data from 3 cohorts (Framingham Heart Study, MESA, Cardiovascular Health Study)
Main finding
As reported in the abstract
Net reclassification improvement, ECG-AI added to PREVENT-HF
1-directional NRI at PREVENT-HF threshold 10%
0.086 to 0.125
1-directional NRI at PREVENT-HF threshold 20%
0.327 to 0.403
Positive composite ECG-AI screen: higher risk of developing heart failure (HF) vs negative screen
10- to 20-fold
Observational data: associations, not effects.
Further findings
| Outcome | ECG-AI LEF | ECG-AI DD | Composite ECG-AI | PREVENT-HF | |
|---|---|---|---|---|---|
| Positive screening rates | 2.9% | 11.1% | 11.9% | ||
| Score ≥10% | Not reported in the abstract | Not reported in the abstract | Not reported in the abstract | 25.1% | ECG-AI LEF, ECG-AI DD, Composite ECG-AI: not reported in the abstract |
| Score ≥20% | Not reported in the abstract | Not reported in the abstract | Not reported in the abstract | 5.8% | ECG-AI LEF, ECG-AI DD, Composite ECG-AI: not reported in the abstract |
Incident HF and death
- Incident HF
- 7.7%
- Death
- 15.1%
Original publication
Predicting Heart Failure From 12-Lead ECGs Using AI: A HeartShare/AMP-HF Pooled Cohort Analysis. J Am Coll Cardiol. Published 8 November 2025. doi:10. 1016/ j. jacc. 2025. 10. 065
Summarised from the published abstract; every number is checked against it