- The FDA cleared the first artificial intelligence tool to detect hidden structural heart disease during routine ECGs.
- EchoNext is a heart screening designed to detect underlying disease before symptoms present.
- The new tool is credited with spotting severe, undiagnosed heart failure in a 45-year-old man who ultimately received a successful heart transplant.
Cardiovascular disease, the
EchoNext is an artificial intelligence tool that reads a standard electrocardiogram (ECG) and flags those at high risk for structural heart disease, according to its maker, Pathway Labs.
The company said that it’s the first AI tool approved by the FDA to detect this kind of hidden heart disease from an electrocardiogram, and that the clearance covers six indications.
In June,
Larger trials are underway, and Pathway Labs has raised $8.5 million to expand into more health systems, according to the company.
EchoNext was built by researchers at NewYork-Presbyterian and Columbia University, led by Pathway Labs’ founder and CEO Pierre Elias, MD, an assistant professor of Medicine in the Department of Biomedical Informatics at Columbia University Data Science Institute.
“We don’t have a screening test for the most common cause of death in the world, which is most forms of cardiovascular disease,” Elias said in a video released by NewYork-Presbyterian.
“So we asked ourselves, could we take a cheap and ubiquitous test, and using AI, turn it into a screening test? And it turns out, we can do exactly that,” Elias continued.
Structural heart disease means there’s a problem with the heart’s valves, chambers, or walls, and it keeps the heart from working the way it should.
Unlike a heart attack, which happens when a blocked artery cuts off blood flow, structural problems are mechanical and don’t always involve blockages.
A valve can stiffen and fail to open fully, forcing the heart to work harder to pump blood through. A valve can leak, allowing blood to flow backward instead of forward. Or the heart muscle can grow too thick or too weak to pump efficiently.
“The heart is a house, and the [ECG] tells us about the electricity. Sometimes, if the electricity demonstrates that there may be issues, we may need additional tests,” said Rachel Bond, MD, co-chair of the Women and Children Committee for the Association of Black Cardiologists.
ECGs have long been used to detect irregular rhythms and heart attacks, not structural problems, which require an echo to diagnose.
EchoNext is designed to bridge that gap. According to Pathway Labs, the tool analyzes the ECG waveform to flag those who should proceed to an echocardiogram. The company said the model was trained on more than 700,000 paired ECG and echocardiogram records across the NewYork-Presbyterian health system.
In a 2025 study published in
EchoNext is cleared to flag six conditions. Pathway Labs has shared five of them:
These conditions often build quietly. According to the Nature study, symptoms of the most common forms, like heart failure and valve disease, can stem from many causes and often surface only late in the disease’s course, after the options for treating it have narrowed.
EchoNext could catch disease earlier; Elias said it spots conditions “the human eye can’t.” It could also help rule out unnecessary tests — something Bond said providers can underestimate.
“The cost of these tests becomes very much of a strain on our patient populations just in terms of copayment and things of that nature,” Bond said.
When a low-risk ECG reading helps confirm that an echocardiogram isn’t necessary, it can also spare a patient the hours off work and the child care and transportation needed for one more appointment, she said.
Diagnosis often depends on whether a patient is referred for an echocardiogram in the first place, and who gets those referrals is uneven, said Aaron Horne, MD, a cardiologist and co-chair of the Association of Black Cardiologists’ Structural Heart Disease Task Force Program.
Fewer African American patients receive valve-replacement procedures, including the less-invasive transcatheter approach, and the treatment gap persists even though outcomes after treatment are often similar by race in some studies, according to a 2020 article published in the
Pathway Labs is partnering with a platform used by physicians called OpenEvidence to put EchoNext in front of doctors quickly.
“FDA-approved AI shouldn’t sit siloed in the ivory tower while patients wait years for it to reach them,” Travis Zack, OpenEvidence’s chief medical officer, said in a news release. “Putting EchoNext on OpenEvidence means a breakthrough in heart disease detection is available everywhere care happens, from major hospitals to community practices.”
Bond said that AI should be one input when determining a diagnosis, but not the final word.
“AI is a tool, and that’s what we should be using it for as one singular tool. It cannot be the end-all,” she said. “When we’re thinking about patient care… we also have to go back to the basics, which is taking a detailed history, factoring in a lot of things that AI may have limitations of.”
Bond further warned about bias in the training data, paired with bias in medicine.
“AI is only as good as the data it’s trained on, and if the training population doesn’t adequately represent populations, specifically individuals of color, specifically Black patients, women, or younger adults, or really underserved communities, the algorithm is going to perform differently across those populations. So we are scrutinizing all of these tools,” she said.
On that question, the Nature study reported no clinically relevant differences in the model’s performance by race, ethnicity, or sex. Across racial and ethnic groups in the study’s main test set, the tool scored about 85 out of 100 on a standard measure of diagnostic accuracy — roughly the same for Black, White, Hispanic, and Asian patients, according to the study.
Healthline contacted the authors of the Nature study, but they couldn’t be reached for comment.
“If we already know that there are certain people starting at a deficit, my biggest concern is that when we bring these new technologies… you’re further exacerbating the disparity that exists,” he said.
Horne added that early detection is critical. Studies show that untreated, severe symptomatic aortic valve narrowing carries a survival rate of around 50% at 2 years after symptoms begin.
Prevention is also crucial. According to
Experts say that adopting heart-healthy lifestyle habits, such as following the
Leave a Reply