MORGANTOWN, W.Va. – A recent study led by Partho Sengupta, M.D., chief of Cardiology and chair of cardiac innovation at the WVU Heart and Vascular Institute, suggests that signal processed electrocardiography (ECG) when combined with artificial intelligence can detect cardiac disease in its early stages.
    

Partho Sengupta, M.D.
Partho Sengupta, M.D.

The study evaluated the use of ECG for predicting abnormal myocardial relaxation, an early indicator of various forms of heart disease, including left ventricular diastolic dysfunction (LVDD), which occurs when the ventricles cannot fill properly. LVDD is responsible for symptoms associated with almost any cardiovascular disease, including high blood pressure, diabetes, heart valve disease, ischemia, and reduced systolic function. It is a common starting point for all cardiac diseases that often leads to heart failure if unattended at an early stage.

“The heart has a pumping phase when it ejects blood and a relaxation phase when it fills. Our hypothesis was that if you could look into subtle changes in electrical signals during the relaxation phase and use machine learning to extract these subtle anomalies, could we predict the presence of early muscle dysfunction that would be normally diagnosed only using an echocardiogram,” Dr. Sengupta said. “The study included two cohorts of patients, one from New York and another one in West Virginia, and found that the technique had robust diagnostic value for predicting muscle relaxation anomalies.”

Typically, echocardiography is used to diagnose LVDD, but it is not used for screening every patient because of expense. By using a novel signal-processed surface ECG algorithm, researchers were able to identify a potential screening method for earlier diagnosis of LVDD.

“Ultrasound cannot be done on everyone because it is expensive and can only be used when clinically indicated,” Sengupta said. “This research looks at how to fill the gap in screening for disease.”

The study, titled “Prediction of Abnormal Myocardial Relaxation from Signal Processed Surface ECG,” appeared in the April 17 issue of the Journal of the American College of Cardiology (JACC), one of the highest-ranked peer reviewed journals in cardiovascular medicine. It is published with editorial commentary primarily authored by Jeroen J. Bax, M.D., Ph.D., president of the European Society of Cardiology. The online version of the study includes an audio summary by Valentin Fuster, M.D., Ph.D., JACC editor in chief.

Co-authors of the study include Hemant Kulkarni, M.D., from M&H Research, L.L.C., in San Antonio, Texas, and Jagat Narula, M.D., Ph.D., from the Department of Cardiology at the Icahn School of Medicine at Mount Sinai University in New York City.