Abstract / Summary
Background: Atrial fibrillation (AF) is a progressive condition that is often undiagnosed until adverse events occur. Traditional risk scores provide limited value for predicting short-term latent AF. Artificial intelligence (AI)-enabled electrocardiogram (ECG) algorithms enable early risk stratification from sinus rhythm ECGs, however their clinical utility remains unvalidated in randomized controlled trials (RCTs).
Objective: To evaluate an AI-ECG-guided strategy for targeted ambulatory Holter monitoring to enhance the early detection of AF or pre-AF atrial findings (PAAFs, defined as premature atrial complexes [PACs] > 500 beats per 24 hours, consecutive short runs of PACs > 20 beats, or non-sustained AF or atrial flutter) compared with usual care.
Design: The LATENT trial is a prospective, open-label, two-arm RCT (NCT06847932) that will recruit 14,726 participants undergoing routine ECGs at a tertiary medical center in Taiwan. Participants will be randomized 1:1 to the intervention group (AI-ECG risk stratification) or control group (usual care). In the intervention group, high-risk individuals identified by the AI-ECG model will be invited to undergo up to 7 days of ambulatory Holter monitoring. In the control group, AI-ECG analysis will be retrospectively applied after study completion to identify those at high risk. The primary endpoint is newly detected AF or PAAFs within 90 days.
Summary: This trial will evaluate the clinical utility of an AI-ECG-guided risk stratification strategy combined with targeted ambulatory Holter monitoring for the early detection of AF or PAAFs in a real-world hospital setting. The findings will inform future strategies for integrating AI-ECG technology to support early AF detection and prevention.