An artificial intelligence framework for improving predictive performance of left ventricular hypertrophy using electrocardiography

Left ventricular hypertrophy (LVH) is defined as an increase in left ventricular mass (LVM) associated with structural changes of myocardium. LVH is a risk factor in cardiovascular disease and clinical diagnosis is important. Electrocardiography (ECG) is the easiest, non-invasive, and economical diagnostic method. Diagnostic methods using ECG have been studied, but performance has been low. Recently, deep learning algorithms using big data and convolution layers have been in the spotlight. In this study, we developed a deep learning-based AI algorithms that can predict LVH in consideration of gender. Both algorithms had some linear relationship between the target value and the predicted value of the model, so the possibility of prediction could be confirmed.