Arrhythmias & electrophysiology
Prediction of Atrial Fibrillation From the ECG in the Community Using Deep Learning: A Multinational Study
Authors' conclusion
In multinational cohort studies, the single-input ECG-AF deep neural network model demonstrated good performance in predicting AF and other cardiovascular outcomes, comparable to a multivariable clinical risk score, with improved performance when combined.
Study context
- Population
- Community cohorts free of atrial fibrillation (AF): Framingham Heart Study (FHS), UK Biobank, ELSA-Brasil3 cohorts
- Comparison
- ECG-AF deep-learning model vs CHARGE-AF clinical risk score
- Outcome
- Discrimination for incident AF (AUC)
Methodological detailsmultinational cohorts, deep neural network, external validation in 3 cohorts, FHS n = 10 097; UK Biobank n = 49 280; ELSA-Brasil n = 12 284, Model refined in 60% of FHS samples; tested in remaining FHS, UK Biobank, ELSA-Brasil
- Multinational cohorts
- Deep neural network
- External validation in 3 cohorts
- FHS n = 10 097; UK Biobank n = 49 280; ELSA-Brasil n = 12 284
- Model refined in 60% of FHS samples; tested in remaining FHS, UK Biobank, ELSA-Brasil
Performance
As reported in the abstract
Incident AF discrimination (FHS)
| Test or model | AUC |
|---|---|
| ECG-AF | 0.82 (95% CI 0.80-0.84) |
| CHARGE-AF | 0.83 (95% CI 0.81-0.85) |
| ECG-AF + CHARGE-AF | 0.85 (95% CI 0.83-0.87) |
AUC
ECG-AF0.82 (95% CI 0.80-0.84)
CHARGE-AF0.83 (95% CI 0.81-0.85)
ECG-AF + CHARGE-AF0.85 (95% CI 0.83-0.87)
Further findings
Incident AF prediction, UK Biobank and ELSA-Brasil
- Combining ECG-AF and CHARGE improved prediction
HF, MI, stroke, all-cause mortality
- Higher ECG-AF scores associated with increased risks in all 3 cohorts
Original publication
Prediction of Atrial Fibrillation From the ECG in the Community Using Deep Learning: A Multinational Study. Circ Arrhythm Electrophysiol. Published 30 September 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →