Left ventricular systolic dysfunction screening in muscular dystrophies using deep learning-based electrocardiogram...

Area of Science:

• Cardiology
• Medical Artificial Intelligence
• Neuromuscular Disorders

Background:

• Routine echocardiography is recommended for muscular dystrophy patients to detect left ventricular systolic dysfunction (LVSD).
• Physical limitations in muscular dystrophy patients often challenge standard echocardiographic monitoring.
• This study investigates the utility of AI-ECG for detecting and predicting LVSD in this population.

Purpose of the Study:

• To evaluate the effectiveness of artificial intelligence-based electrocardiogram interpretation (AI-ECG) in detecting left ventricular systolic dysfunction (LVSD) in patients with muscular dystrophy.
• To assess the predictive value of AI-ECG for new-onset LVSD in muscular dystrophy patients.

Main Methods:

• A convolutional neural network (CNN) was trained on a large dataset to detect LVSD.
• The AI-ECG model was tested on a cohort of patients with Duchenne (DMD), Becker (BMD), limb-girdle muscular dystrophy (LGMD), and myotonic dystrophy (MD).
• Cox proportional hazards models were used to evaluate the predictive capability of AI-ECG for incident LVSD.

Main Results:

• The AI-ECG model demonstrated an AUROC of 0.83 for detecting LVSD in the muscular dystrophy test set.
• The model achieved a sensitivity of 0.87 and a negative predictive value (NPV) of 0.91.
• AI-ECG predicted new-onset LVSD with an AUROC of 0.72, with AI-ECG probability being a significant predictor.

Conclusions:

• AI-ECG serves as a non-invasive and accessible tool for detecting LVSD in muscular dystrophy patients.
• AI-ECG can aid in risk stratification and potentially serve as an alternative to routine echocardiography.
• The findings suggest AI-ECG may predict new-onset LVSD, facilitating earlier therapeutic interventions.