Detection of Hypertrophic Cardiomyopathy on Electrocardiogram Using Artificial Intelligence

Area of Science:

• Cardiology
• Medical Artificial Intelligence
• Diagnostic Tools

Background:

• Hypertrophic cardiomyopathy (HCM) is a significant cause of morbidity and mortality, particularly sudden cardiac death in young individuals.
• The condition is estimated to affect 1 in 500 people, with many cases remaining undiagnosed.
• Improved screening methods, such as electrocardiogram (ECG) analysis, could enhance early detection and diagnosis of HCM.

Purpose of the Study:

• To evaluate the accuracy of an artificial intelligence (AI) device in detecting hypertrophic cardiomyopathy (HCM) using a standard 12-lead electrocardiogram.
• To assess the potential of AI in augmenting the diagnostic capabilities for HCM.

Main Methods:

• A deep learning-based AI device was utilized, providing a binary output: 'HCM suspected' or 'not suspected'.
• The study included a dataset of 293 HCM-positive and 2912 HCM-negative cases, identified through chart review across three hospitals.
• The AI device processed 291 (99.3%) HCM-positive and 2905 (99.8%) HCM-negative cases.

Main Results:

• The AI device achieved a sensitivity of 68.4% and a specificity of 99.1% for HCM detection.
• The area under the curve (AUC) was 0.975, indicating strong discriminatory performance.
• With an assumed prevalence of 0.2% (1 in 500), the positive predictive value was 13.7% and the negative predictive value was 99.9%.

Conclusions:

• The AI device demonstrated good performance in identifying hypertrophic cardiomyopathy from 12-lead electrocardiograms.
• When used in conjunction with clinical expertise, this AI tool has the potential to improve the detection and diagnosis of HCM.