Arrhythmias & electrophysiologyCohort study
Artificial intelligence–derived electrocardiographic aging and risk of atrial fibrillation: a multi-national study
- Retrospective cohort study
- multi-national
- AI-ECG age prediction model development and validation
Model developed on 1 533 042 electrocardiogram (ECGs) from 689 639 participants; validated on 6 multinational datasets (737 133 ECGs, 330 794 participants)
Authors' conclusion
The AI-derived ECG aging was associated with the risk of new- and early-onset AF, suggesting its potential utility to identify individuals for AF prevention across diverse populations.
Study context
- Population
- Adults with 12-lead ECG in two South Korean, one UK and one US cohorttwo South Korean, one UK, one US cohort
- Exposure
- ECG-aged vs normal
- Outcome
- New-onset AF (and early-onset AF)
Main finding
As reported in the abstract
New-onset AF, ECG-aged vs normal
ECG-aged vs NormalHazard ratio (95% CI)
South Korean cohort 1
2.50(2.24–2.78)
South Korean cohort 2
1.89(1.46–2.43)
UK cohort
1.90(1.55–2.33)
US cohort
1.76(1.67–1.86)
Observational data: associations, not effects.
Further findings
| Outcome | OR (95% CI) | |
|---|---|---|
| Early-onset AF, ECG-aged vs normal — ECG-aged vs Normal · South Korean cohort 1 | 2.89 (2.47–3.37) | |
| Early-onset AF, ECG-aged vs normal — ECG-aged vs Normal · South Korean cohort 2 | 1.94 (1.39–2.70) | |
| Early-onset AF, ECG-aged vs normal — ECG-aged vs Normal · UK cohort | 1.58 (1.06–2.35) | |
| Early-onset AF, ECG-aged vs normal — ECG-aged vs Normal · US cohort | 1.79 (1.62–1.97) | |
| 0.20.5125Odds ratio (log scale) |
Mean AI-ECG age and age gap by cohort
- South Korean cohort 1: AI-ECG age 51.9 (SD 16.2) y; age gap .0 (SD 6.8) y
- South Korean cohort 2: AI-ECG age 47.4 (SD 12.5) y; age gap −.1 (SD 6.0) y
- UK cohort: AI-ECG age 68.4 (SD 7.8) y; age gap 4.7 (SD 8.7) y
- US cohort: AI-ECG age 56.7 (SD 14.6) y; age gap −1.4 (SD 8.9) y
Original publication
Artificial intelligence–derived electrocardiographic aging and risk of atrial fibrillation: a multi-national study. Eur Heart J. Published 29 November 2024. doi:10.1093/eurheartj/ehae790
Summarised from the published abstract; every number is checked against it