"It beats cardiologists": This is how artificial intelligence reveals heart diseases invisible to a standard ECG.

An electrocardiogram , a rapid and inexpensive test, could become an advanced key to the early diagnosis of structural heart disease. This is thanks to Artificial Intelligence (AI), and specifically, a tool called EchoNext , developed by researchers at Columbia University and New York-Presbyterian, recently described in the journal Nature. With the help of AI, data from a simple ECG can now reveal complex conditions—such as valvular heart disease, congenital heart disease, or cardiac thickening—that until now could only be diagnosed with more sophisticated tests like echocardiograms.

How does it work?

EchoNext analyzes waveforms to identify at-risk patients and direct them to targeted tests. In tests conducted so far, EchoNext has outperformed cardiologists in diagnostic accuracy, even those who use AI to read ECGs. The system, explains lead author Pierre Elias, "uses the most cost-effective test to uncover pathologies that are normally undetected. We believe that AI-enhanced ECGs could open a new paradigm in cardiac screening."

The model was trained on over 1.2 million ECG-echocardiogram pairs from a sample of approximately 230,000 patients. During the clinical validation phase, EchoNext demonstrated high accuracy in identifying structural heart conditions such as heart failure due to cardiomyopathy, valvular heart disease, pulmonary hypertension, and myocardial thickening. In a direct comparison with 113 cardiologists on 3,200 ECGs, EchoNext correctly identified 77% of pathological cases, while cardiologists—even with AI support—only achieved 64%. This is a significant difference, especially during screening. To evaluate the tool's real-world impact, the team then analyzed nearly 85,000 patients who had undergone an ECG but never had an echocardiogram. EchoNext flagged over 7,500 people (9%) as being at high risk for undiagnosed structural heart disease. “We could transform the 400 million ECGs performed each year worldwide into opportunities to identify heart conditions before they become serious,” Elias explains, “and offer lifesaving treatments at the right time.”

The expert: "Time gained for treatment"

"This study has essentially transformed a useful tool like the ECG , which boasts 122 years of history, into a support that, thanks to Artificial Intelligence, is significantly updated and becomes a particularly effective tool for the cardiologist's work," explains Dr. Rocco Montone , cardiologist at the Gemelli Polyclinic in Rome, to FattoQuotidiano.it . "It allows us to understand, through small alterations present in the electrocardiogram, if there is a more serious malformation that was hidden and direct the patient towards other tests such as the echocardiogram. Let us remember that the ECG records the electrical activity of the heart and can show us arrhythmias, ischemia, past heart attacks , and indirect signs of hypertrophy or cardiac strain. Its limitation is that it does not show the structure of the heart; while the echocardiogram, a more expensive test than the previous one, allows us to see valves, heart chambers, wall thickness, contractility and the presence of structural defects . Understanding which patients deserve to pass at the second level of diagnosis it allows for a much more accurate and analytical screening, avoiding inviting patients to undergo further tests without really needing them”.

A significant advantage in terms of time and economic resources saved and diagnostic accuracy.

Certainly, but not only that. The time saved on the diagnostic instrumental examination benefits the doctor-patient relationship. The specialist, in this case, can devote more time to listening and observing the patient, all essential aspects for establishing a relationship of trust and improving the treatment and healing process.

So isn't Artificial Intelligence at risk of replacing doctors?

"No, definitely not. All aspects related to the humanization of therapies can be enhanced precisely thanks to these tools that speed up and improve intermediate diagnostic steps. Because empathy and connection with the person being treated certainly cannot be replaced by machines. And this should be a stimulus for us doctors to further develop the human dimension of medicine, one of the factors that can improve the positive outcomes of treatments."