Key result
Planned meta-analysis compares AI models with conventional risk scores for predicting incident AF.
Why the study?
The comparative discrimination, calibration, and incremental value of artificial intelligence-based models versus conventional risk scores for predicting incident atrial fibrillation in adults warrant systematic synthesis.
Do artificial intelligence-based models improve the discrimination, calibration, and incremental value for predicting incident atrial fibrillation compared to conventional risk scores in adults?
Meta-Analysis
Do artificial intelligence-based models improve the discrimination, calibration, and incremental value for predicting incident atrial fibrillation compared to conventional risk scores in adults?
This protocol outlines a planned systematic review and network meta-analysis to evaluate whether AI-based prediction models outperform conventional clinical risk scores in predicting incident atrial fibrillation.
Answers are grounded in this paper and cite their sources.
Protocol for meta-analysis of AI models versus conventional scores in incident AF prediction; leaves open comparative performance pending completion.
Full protocol (version 1.2) for a systematic review and meta-analysis comparing the discrimination, calibration and incremental value of artificial intelligence-based models (AI-ECG, machine learning on electronic health records) with conventional risk scores (CHARGE-AF, C2HEST, CHA2DS2-VASc) for predicting incident atrial fibrillation in adults. Includes the search strategy and statistical analysis plan, including pairwise and network meta-analysis of C-statistics. Review start date: 1 September 2026; expected completion: 20 November 2026. Prospectively registered: at registration only pilot work (preliminary searches) had been done; the formal search, screening, data extraction and synthesis had not started. PROSPERO registration in preparation.
No takes yet. Share an insight, caveat, or question.
Musa Bin Bashir (2026) conducted a meta-analysis in Incident atrial fibrillation. Artificial intelligence-based models (AI-ECG, machine learning on electronic health records) vs. Conventional risk scores (CHARGE-AF, C2HEST, CHA2DS2-VASc) was evaluated on Discrimination (C-statistics), calibration, and incremental value for predicting incident atrial fibrillation. Artificial intelligence-based models will be compared with conventional risk scores for predicting incident atrial fibrillation in a planned meta-analysis.
Synapse has enriched 10 related papers. Consider them for comparative context: