Arrhythmias & electrophysiology
Near-term prediction of sustained ventricular arrhythmias applying artificial intelligence to single-lead ambulatory electrocardiogram
Authors' conclusion
A novel deep learning model utilizing dynamic single-lead ambulatory ECGs accurately identifies patients at near-term risk of ventricular arrhythmias. It also uncovers an early depolarization pattern as a potential determinant of ventricular arrhythmias events.
Study context
- Population
- 14-day ambulatory electrocardiogram (ECG) recordings from six countries
- Participants
- 247,254 recordings
- Comparison
- Deep learning model on first 24 h of single-lead ambulatory ECG for prediction of sustained ventricular tachycardia (VT) in subsequent 13 days
- Internal validationn = 43,580
- External validationn = 20,497
- Outcome
- Sustained VT occurrence
- Follow-up
- Subsequent 13 days
Methodological detailsdeep learning model, internal and external validation, 6 countries
- Deep learning model
- Internal and external validation
- 6 countries
Performance
As reported in the abstract
Prediction of sustained VT
| Test or model | AUC | Sensitivity at specificity 97.0% |
|---|---|---|
| Internal validation | .957 (95% CI .943-.971) | 70.6% |
| External validation | .948 (95% CI .926-.967) | 66.1% |
AUC
Internal validation.957 (95% CI .943-.971)
External validation.948 (95% CI .926-.967)
Sensitivity at specificity 97.0%
Internal validation70.6%
External validation66.1%
Further findings
Prediction of rapid sustained VT (≥180 b.p.m.)
Sustained ventricular arrhythmias, all recordings
- Recordings with sustained ventricular arrhythmias
- 1104 (.5%)
Prediction of VT degenerating into VF
- VT that degenerated into VF accurately predicted
- 90.0%
Saliency mapping
- PVC burden and early depolarization time suggested as predictors for VT
Original publication
Near-term prediction of sustained ventricular arrhythmias applying artificial intelligence to single-lead ambulatory electrocardiogram. Eur Heart J. Published 30 March 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →