Arrhythmias & electrophysiologyCohort study
PVCuRe: A machine-learning–based tool to predict left ventricular systolic function recovery in patients undergoing ablation for premature ventricular contraction
- retrospective multicenter study
- development and external validation of a machine-learning model
- 3 international referral centers
- external validation cohort from 3 additional centers
Authors' conclusion
This ML-based tool, built on widely available variables, accurately estimates the probability of LVEF recovery after PVC ablation and may support clinical decision making and patient counseling.
Study context
- Population
- Patients with left ventricular ejection fraction (LVEF) <50% who underwent successful PVC ablationpatients with LVEF <50% who underwent successful PVC ablation at 3 international referral centers
- Participants
- 256 patients
- Exposure
- Machine-learning (ML) model to predict LVEF recovery after PVC ablation, assessed with 10-fold cross-validation and receiver operating characteristic curve analysis, with external validation in an independent cohort
Further findings
Discrimination and calibration of the Random Forest model (internal test set)
- area under the curve
- 0.88 (95% confidence interval = 0.79-0.98)
- Hosmer-Lemeshow P
- P = 0.562
- Random Forest model showed the best performance of 5 ML models
Discrimination in external validation
- area under the curve
- 0.83, 95% confidence interval = 0.72-0.95
Key predictors
- baseline PVC burden, QRS duration in sinus rhythm, and preprocedural LVEF
Original publication
PVCuRe: A machine-learning–based tool to predict left ventricular systolic function recovery in patients undergoing ablation for premature ventricular contraction. Heart Rhythm. Published 4 September 2026. doi:10.1016/j.hrthm.2026.08.058
Summarised from the published abstract; every number is checked against it