Artificial intelligence-enhanced electrocardiogram for arrhythmogenic right ventricular cardiomyopathy detection

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Electrocardiogram (ECG) abnormalities are key indicators of arrhythmogenic right ventricular cardiomyopathy (ARVC).
• Developing advanced diagnostic tools for ARVC is crucial for early detection and management.

Purpose of the Study:

• To develop and evaluate an artificial intelligence (AI)-enhanced ECG model for identifying patients with ARVC.
• To assess the potential of AI-ECG as a non-invasive disease-detection tool for ARVC.

Main Methods:

• A convolutional neural network (CNN) was designed to analyze 12-lead ECGs for ARVC detection.
• The model was trained and validated using ECGs from patients meeting 2010 ARVC task force criteria and confirmed genetic variants, matched with control ECGs.
• The dataset included 77 ARVC cases (various genetic variants) and a large control group.

Main Results:

• The AI-ECG model demonstrated a sensitivity of 77.3% and specificity of 62.9%.
• The negative predictive value was high at 99.4%, indicating strong performance in excluding ARVC.
• The area under the curve (AUC) for rhythm and median beat ECGs was 0.75 and 0.76, respectively.

Conclusions:

• The AI-ECG model shows potential as a biomarker for ARVC, particularly in excluding the disease.
• Further refinement and validation of the algorithm using larger, multicenter cohorts are recommended.
• AI-enhanced ECG analysis represents a promising avenue for ARVC diagnostics.