AI-Enabled ECG Software Predicts One-Year Atrial Fibrillation Risk

Atrial fibrillation (AF) is the most common cardiac arrhythmia and is associated with increased risks of stroke, heart failure, and death. Detection remains challenging because AF is often asymptomatic and intermittent, driving interest in risk stratification using routine electrocardiogram (ECG) data. A new artificial intelligence (AI) model applied to standard 12-lead ECGs has demonstrated validated performance in predicting one-year AF risk.

Tempus AI (Chicago, IL, USA) has announced a successful multi‑site validation of its ECG‑AF software, with results published in Heart Rhythm under the title “Multi‑Center Validation of an Artificial Intelligence‑Enabled ECG Model to Predict 1‑Year Risk of Atrial Fibrillation or Flutter.” The software received U.S. Food and Drug Administration (FDA) clearance in 2024 for predicting the one‑year risk of AF or atrial flutter. The multi‑center evaluation spanned three geographically distinct clinical sites and assessed performance in older adults who did not have known AF at baseline. The resulting data are described as surpassing pre‑specified performance thresholds and supporting the device’s clearance.

The ECG‑AF software analyzes resting 12‑lead ECG recordings to generate an AI-derived risk score indicating the likelihood that a patient will experience AF or atrial flutter within 12 months. It is intended for use in healthcare facilities on patients aged 65 years or older without pre‑existing or concurrent AF or atrial flutter and without a pacemaker or implantable cardioverter defibrillator, and not within 30 days of cardiac surgery. The tool analyzes ECG data only and should be interpreted alongside the original ECG, other tests, symptoms, and clinical history. It is not validated for monitoring, ruling out AF, or as the sole basis for diagnosis or treatment.

In the study, investigators aggregated ECG data and conducted manual chart reviews to select eligible participants and ascertain outcomes. Endpoints were defined as a new AF diagnosis within one year or one year of AF‑free follow‑up. Among 4,017 evaluated patients, the AI-derived risk score exceeded the pre‑specified thresholds. The company notes this was the first FDA‑cleared ECG‑AI device in its portfolio aimed at identifying patients at risk for cardiovascular conditions.

“This study marks an important step toward shifting cardiac care from late-stage intervention to early risk detection,” said Brandon Fornwalt, MD, PhD, SVP of Cardiology at Tempus and a coauthor of the study. “The ability of our AI model to consistently predict atrial fibrillation across varied clinical environments highlights its potential as a dependable decision-support tool. We believe this will enable clinicians to surface hidden risks sooner, opening the door to earlier, more targeted diagnosis and care to help minimize serious complications such as stroke and heart failure.” 

Related Links
Tempus AI