An ordinary test found in many doctors’ offices could soon become a powerful way to detect hidden heart problems—thanks to artificial intelligence (AI).
Millions of people around the world suffer from structural heart disease. This includes valve problems, congenital issues, and other defects that affect how the heart functions. But many of these conditions go unnoticed until it’s too late.
Unlike cancer, which has widely used screening tools like mammograms and colonoscopies, there’s no routine or affordable way to screen for most types of heart disease.
Now, researchers from Columbia University and NewYork-Presbyterian have developed a tool called EchoNext that uses AI to analyze results from electrocardiograms (ECGs)—one of the most common heart tests in medicine.
Normally, ECGs are used to detect abnormal rhythms or past heart attacks. They are quick, cheap, and non-invasive. However, doctors have long believed that ECGs can’t show structural problems in the heart.
EchoNext is changing that. The AI system scans ECG data and determines whether a patient needs a follow-up test called an echocardiogram, which uses ultrasound to create images of the heart. These images can reveal more serious problems like valve disease, cardiomyopathy, and pulmonary hypertension.
According to Dr. Pierre Elias, who led the study, EchoNext can spot signs of heart disease that cardiologists may miss when looking at ECGs alone.
In fact, the AI tool outperformed 13 cardiologists in a head-to-head comparison. EchoNext correctly identified 77% of structural heart conditions from ECGs, while the doctors had an accuracy of just 64%.
The AI model was trained using more than 1.2 million ECG and echocardiogram pairs from 230,000 patients. It was tested across four hospital systems and proved highly accurate in spotting a range of heart conditions. EchoNext essentially uses the cheaper test—the ECG—to decide who should get the more expensive and detailed test—the echocardiogram.
To see how EchoNext performs in real-life settings, researchers ran the tool on nearly 85,000 patients who had received an ECG but never had an echocardiogram. The AI flagged more than 7,500 people as high risk.
About 55% of those individuals later got an echocardiogram, and nearly 75% of them were diagnosed with structural heart disease. That’s twice the usual detection rate.
If every high-risk patient had followed up with an echocardiogram, doctors estimate they could have diagnosed an extra 2,000 people with serious heart problems.
Dr. Elias believes this technology could completely change how we screen for heart disease. Around 400 million ECGs are performed worldwide each year. If EchoNext were widely adopted, each of those ECGs could become an opportunity to catch heart disease early and possibly save lives.
The researchers are now running clinical trials at eight emergency departments to see how EchoNext works in fast-paced hospital settings. They’ve also released a deidentified dataset so other researchers and hospitals can continue improving AI tools for heart disease screening.
In short, this breakthrough could bring heart disease screening to millions of people using a test that’s already widely available—just made smarter by AI.
If you care about heart health, please read studies that Manganese can help clear arterial plaques and treat heart disease and Aspirin use linked to heart failure.
For more about heart health, please read studies about the blood thinner drug that can prevent strokes in people with hidden heart issues and new guidelines on daily aspirin for heart attack and stroke prevention.
The study is published in Nature.
Copyright © 2025 Knowridge Science Report. All rights reserved.
