An AI-ECG-Based Approach for Dynamic Assessment of Heart Failure Risk and Myocardial Recovery Following Atrial Fibrillation Ablation

DYNAMIC-AF HF Study: An AI-ECG-Based Approach for Dynamic Assessment of Heart Failure Risk and Myocardial Recovery Following Atrial Fibrillation Ablation

Background Artificial intelligence-enabled electrocardiography (AI-ECG) has emerged as a promising digital biomarker for detecting latent myocardial dysfunction and predicting cardiovascular risk. However, whether serial AI-derived risk estimates reflect myocardial recovery following therapeutic intervention remains unknown.

Objective The DYNAMIC-AF HF Study aims to evaluate longitudinal changes in AI-ECG-derived heart failure (HF) risk after catheter ablation in patients with atrial fibrillation (AF) and heart failure with mildly reduced ejection fraction (HFmrEF), and to determine their association with conventional markers of reverse remodeling.

Methods The DYNAMIC-AF HF Study is a prospective multicenter observational cohort study enrolling 1,000 patients with symptomatic AF and HFmrEF undergoing first-time catheter ablation. Eligible participants must have a left ventricular ejection fraction of 41-49% and at least one predefined HF-related feature suggestive of latent myocardial dysfunction. Serial 12-lead electrocardiograms, echocardiography, biomarker assessments, and clinical follow-up will be performed at baseline and at 3, 6, and 12 months. AI-based ECG analysis will generate continuous HF-risk scores, enabling construction of longitudinal AI-derived HF risk trajectories. The primary endpoint is the change in AI-derived HF risk from baseline to 12 months. Secondary endpoints include changes in left ventricular ejection fraction, global longitudinal strain, N-terminal pro-B-type natriuretic peptide levels, AF recurrence, HF hospitalization, and mortality.

Conclusions This study will evaluate whether serial AI-ECG assessment can serve as a dynamic digital biomarker of myocardial recovery following AF ablation and support future AI-enabled monitoring and clinical decision-support strategies in cardiovascular care.

Inclusion Criteria:

• Age ≥18 years

• Symptomatic paroxysmal or persistent atrial fibrillation

• Scheduled for first-time catheter ablation

• Left ventricular ejection fraction between 41% and 49% measured by transthoracic echocardiography within 3 months before ablation

• At least one analyzable sinus rhythm 12-lead electrocardiogram before ablation

• At least one latent heart failure substrate feature:

  • Elevated NT-proBNP
  • Increased left atrial volume index (>34 mL/m²)
  • Average E/e' ≥14
  • Reduced global longitudinal strain
  • Mild pulmonary hypertension
  • Exertional intolerance suggestive of early heart failure

• Ability to provide written informed consent

Exclusion Criteria:

• Left ventricular ejection fraction ≤40%
• Previous atrial fibrillation catheter ablation
• Significant valvular heart disease requiring intervention
• Hypertrophic or infiltrative cardiomyopathy
• Acute coronary syndrome, coronary revascularization, or myocarditis within 3 months
• Severe renal dysfunction requiring dialysis
• Active malignancy with life expectancy <1 year
• Inadequate electrocardiographic or echocardiographic image quality
• Pregnancy
• Inability or unwillingness to provide written informed consent