Valvular & structural heart diseaseCohort study
Generalisable artificial intelligence ECG trained on public data for outcome prediction after transcatheter aortic valve replacement
- developed and externally validated a generalisable Artificial intelligence ECG (AI-ECG) model
- transformer-based
- trained exclusively on publicly available data
- prospective cohort
Authors' conclusion
An AI-ECG model trained solely on publicly available data provides accurate and generalisable prediction of outcomes after TAVR using a single ECG. This approach demonstrates the feasibility of transparent, shareable AI models for cardiovascular risk stratification and supports a general-to-specific paradigm for clinical deployment.
Study context
- Population
- Patients undergoing transcatheter aortic valve replacement (TAVR); additionally a surgical aortic valve replacement cohortprospective TAVR validation cohort, model trained on 341,151 publicly available ECGs
- Participants
- 439 patients
Main finding
As reported in the abstract
30-day mortality prediction (AUC), single pre-procedural ECG · 30-day
AUC
0.85 (95% CI 0.78 to 0.92)
Observational data: associations, not effects.
Further findings
1-year mortality prediction (AUC)
- AUC
- 0.74 (95% CI 0.67 to 0.80)
- TAVR cohort
AI-ECG risk score and 1-year mortality
- adjusted OR
- 1.70
- Independently associated with major adverse cardiac events and major renal events
- multivariable regression
Predictive performance in surgical cohort
- Predictive performance was consistent in an independent surgical cohort
High-risk ECG features
- atrial fibrillation, prolonged QT interval, ST-T abnormalities and low voltage
Original publication
Generalisable artificial intelligence ECG trained on public data for outcome prediction after transcatheter aortic valve replacement. Heart. Published 28 May 2026. doi:10.1136/heartjnl-2025-327462
Summarised from the published abstract; every number is checked against it