Detection of Left Atrial Myopathy Using Artificial Intelligence-Enabled Electrocardiography

Area of Science:

• Cardiology
• Artificial Intelligence in Medicine
• Biomedical Engineering

Background:

• Left atrial (LA) myopathy is prevalent in heart failure with preserved ejection fraction (HFpEF), often preceding atrial fibrillation (AF).
• Identifying LA myopathy early is crucial for managing HFpEF and preventing AF.
• Current methods for assessing LA myopathy are often invasive or require multiple tests.

Purpose of the Study:

• To evaluate a novel AI-enabled ECG analysis for detecting LA myopathy in HFpEF patients.
• To assess the correlation between AI-derived ECG probability of AF and LA structural/functional abnormalities.
• To determine if AI-ECG can predict incident AF and long-term mortality in HFpEF.

Main Methods:

• A cohort of 613 HFpEF patients underwent AI-enabled ECG, echocardiography, and cardiac catheterization.
• Patients were stratified based on AI-ECG probability of concurrent AF as a marker for LA myopathy.
• Long-term follow-up data were analyzed for incident AF and mortality.

Main Results:

• Higher AI-ECG AF probability correlated with more severe structural heart disease, including left ventricular hypertrophy and larger LA volumes.
• Patients with higher AI-AF probability exhibited elevated cardiac filling and pulmonary artery pressures, and impaired cardiac output reserve.
• Each 10% increase in AI-AF probability significantly increased the risk of new-onset AF (HR 1.31) and all-cause mortality (HR 1.12) in the overall cohort.

Conclusions:

• A novel AI-enabled ECG score can identify underlying LA myopathy in HFpEF patients.
• The AI-ECG effectively detects structural, functional, and hemodynamic abnormalities associated with LA myopathy.
• This AI tool shows promise for predicting long-term AF risk in HFpEF, warranting further clinical investigation.