Abstract
Background
Atrial fibrillation (AF) is frequently intermittent and may therefore remain undetected on a short resting electrocardiogram (ECG). At the same time, a sinus-rhythm ECG contains information on the underlying atrial substrate.Objective
To review conventional and artificial intelligence (AI)-based ECG markers of an atrial risk phenotype and their potential role in targeted AF screening.Materials and methods
Narrative appraisal of current consensus and guideline documents and of clinical, electrophysiological, and population-based studies on P‑wave parameters, AI-ECG, and risk-adapted AF screening.Results
P‑wave duration, interatrial block, P‑wave terminal force in V1, and other parameters are associated with atrial remodelling and future AF. AI models additionally extract high-dimensional patterns from sinus-rhythm ECGs and can predict occult or future AF. Initial prospective data indicate that such risk signals can enrich the diagnostic yield of prolonged rhythm monitoring.Conclusion
The most plausible current clinical application is risk-adapted selection for prolonged rhythm monitoring. A P-wave or AI-based risk score neither establishes a diagnosis of AF nor, by itself, constitutes an indication for treatment.