Arrhythmias & electrophysiology
Machine Learning Integration of Ablation Quality Metrics for Predicting Atrial Fibrillation Recurrence: A Multicenter Study
Authors' conclusion
Integrating ablation quality metrics yielded an validated model that showed strong discrimination for 1-year AF recurrence, outperforming clinical features alone. This approach highlights the prognostic value of intraprocedural quality assessment and may support personalized patient management.
Study context
- Population
- Patients aged 18 to 80 years undergoing their first atrial fibrillation (AF) ablationenrolled between December 2021 and December 2023, external validation n = 239 · External validation centers 239
- Participants
- 1,782 patientsEvents: 294 (16.5%)
- Predictors
- XGBoost model using 7 selected features integrating clinical characteristics with ablation quality metrics
- Predicted outcome
- 1-year AF recurrence
Methodological detailsdevelopment and external validation of a machine learning model, 5-fold cross-validation for internal validation, 14 hospitals from mainland China
- Development and external validation of a machine learning model
- 5-fold cross-validation for internal validation
- 14 hospitals from mainland China
Model performance
As reported in the abstract
AUC
Internal validation
0.918, 95% CI: 0.874-0.962
External validation center 1External validation (2 independent centers)
0.992
External validation center 2External validation (2 independent centers)
0.951
Further findings
| Outcome | Values |
|---|---|
| Feature-combination analyses | Clinical features alone AUC 0.588 · After incorporating ablation quality metrics AUC 0.918 |
| External validation (2 independent centers) | External validation center 1 precision-recall AUC 0.960 · External validation center 2 precision-recall AUC 0.815 |
- External validation (2 independent centers): n = 239
Original publication
Machine Learning Integration of Ablation Quality Metrics for Predicting Atrial Fibrillation Recurrence: A Multicenter Study. JACC Clin Electrophysiol. Published 3 October 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →