Key result
New protocol will compare machine learning against traditional risk scores for predicting AF recurrence.
Why the study?
AF recurrence remains common after catheter ablation, but the comparative performance of machine learning models and traditional clinical risk scores for predicting recurrence remains uncertain.
Do machine learning models improve the prediction of atrial fibrillation recurrence after catheter ablation compared to traditional clinical risk scores?
Systematic Review
Do machine learning models improve the prediction of atrial fibrillation recurrence after catheter ablation compared to traditional clinical risk scores?
This systematic review protocol outlines a planned comparison of machine learning models versus traditional risk scores for predicting atrial fibrillation recurrence after catheter ablation.
Performance variation cautions against preferring machine learning models over traditional scores for post-ablation AF recurrence; leaves comparative validation open.
Background: Atrial fibrillation (AF) recurrence remains common following catheter ablation, affecting approximately 25–40% of patients after the blanking period. Numerous machine learning models and traditional clinical risk scores have been developed to predict recurrence; however, their comparative performance remains uncertain. Objective: To systematically review and compare the predictive performance of machine learning models and traditional clinical risk scores for predicting AF recurrence after catheter ablation. Methods: Electronic searches were conducted in PubMed, Scopus, Web of Science, Europe PMC, ClinicalTrials.gov, and medRxiv, supplemented by citation-network expansion using ResearchRabbit and manual reference screening. Studies evaluating recurrence prediction models or risk scores after AF ablation were included. Data on model characteristics, validation strategy, and discrimination metrics (AUC/C-index) were extracted. Risk of bias will be assessed using PROBAST. Planned synthesis: Quantitative synthesis will compare model performance across modalities including multimodal AI, ECG AI, clinical/EHR AI, imaging/radiomics AI, biomarker AI, and traditional clinical scores using median, mean, and range of reported discrimination metrics
No takes yet. Share an insight, caveat, or question.
Farahat et al. (2026) conducted a systematic review in Atrial fibrillation. Machine learning models vs. Traditional clinical risk scores was evaluated on Predictive performance (discrimination metrics such as AUC/C-index) for atrial fibrillation recurrence. This systematic review protocol outlines a planned quantitative synthesis to compare the predictive performance of machine learning models and traditional risk scores for atrial fibrillation recurrence.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: