Cardiologists develop new AI screening tool for structural heart disease
A research team led by Pierre Elias, MD, a cardiologist with Columbia University Irving Medical Center, has developed a new artificial intelligence (AI) tool that uses electrocardiograms (ECGs) to screen patients for various forms of structural heart disease.
The AI model in question, EchoNext, can detect signs of severe aortic stenosis, mitral regurgitation, heart failure and other critical cardiovascular findings.
“In general, AI models take data and run mathematical operations to find unique patterns that a human may miss,” Elias explained in a February interview. “With EchoNext, we input data from a patient’s ECG so it can be analyzed to identify patients who should be evaluated for structural heart disease. In this way, EchoNext ‘sees’ something a cardiologist can’t.”
One of the standout cases so far involved an 84-year-old woman who was hospitalized after fracturing her femur. She presented with no cardiac symptoms at the time, but an ECG analyzed by EchoNext flagged her for potential signs of heart disease and a consultation with a cardiologist was ordered. The patient ended up undergoing TAVR just more than a week later. Elias and his team shared their experience in JACC: Case Reports.[1]
“To the best of our knowledge, this is the first reported case of an AI-enabled ECG screening algorithm detecting undiagnosed severe aortic stenosis, leading to definitive valve intervention with TAVR,” wrote first author Emily Tat, MD, a cardiology fellow with Columbia, and colleagues. “While multiple prior studies have demonstrated the capability for deep learning models to detect structural heart disease from ECGs, real-world clinical examples showing downstream impact on management and outcomes remain limited.”
The group emphasized that this patient’s eventual diagnosis may have come much later without the assistance of EchoNext.
“From a clinical standpoint, early identification of severe aortic stenosis has meaningful implications,” they wrote.
Click here to read the full study in JACC: Case Reports, an American College of Cardiology journal.
The team behind EchoNext appears to be just getting started; a randomized trial is on the schedule for 2026.
Michael has more than 19 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.