A team at Mayo Clinic has built an artificial intelligence model that can flag a serious, often silent heart obstruction from the kind of ultrasound video already recorded in most clinics. The condition is hypertrophic cardiomyopathy, or HCM, and it is the leading cause of sudden cardiac death in young athletes.
The model reads standard two-dimensional echocardiogram clips and looks for signs of left ventricular outflow tract obstruction, the narrowing that traps blood inside a thickened heart. Crucially, it does not require Doppler imaging, the specialist add-on that only trained sonographers can capture reliably. The research was published in Circulation: Cardiovascular Imaging in August 2026.
What HCM is and why it hides
Hypertrophic cardiomyopathy is a genetic disease in which parts of the heart muscle grow abnormally thick. Cleveland Clinic describes it as affecting roughly 1 in 200 to 1 in 500 people, and notes that many cases go undiagnosed. Shortness of breath during activity is the most common symptom, but chest discomfort, palpitations, dizziness and fainting can also appear.
The obstructive form of HCM is the more common one, according to Cleveland Clinic. That is LVOT obstruction, where the thickened muscle or the nearby mitral valve gets in the way of blood leaving the heart, and it is what turns HCM from a background finding into a condition that can cause collapse during exertion.
Also read: A simple shift in what you eat every day may lower your colorectal cancer risk
For young athletes, the stakes are stark. Cleveland Clinic calls HCM the most common cause of sudden cardiac death in athletes under 35.
What the AI does differently
Confirming an outflow obstruction usually means Doppler echocardiography, a technique that measures blood-flow speed and depends on precise beam alignment and a skilled operator. Where that expertise is not on hand, the obstruction can be missed on a routine scan.
The Mayo team trained a deep-learning model on echocardiogram videos from 1,833 patients seen at Mayo Clinic. It was then tested on 275 additional Mayo patients and externally validated on 46 patients from a South Korean hospital, according to the detailed writeup at News-Medical. The model combines information from three standard ultrasound views rather than relying on a single angle.
In a preprint of the work, posted on medRxiv, the multi-view fusion approach reached an AUROC of 0.84 on the external Korean cohort, a substantial jump over single-view baselines. In a subset of scans, the model identified obstruction more accurately than two expert echocardiographers reviewing the same non-Doppler images.
Also read: Nearly 8 in 10 young adults may already have early signs of cardiovascular damage
What routine detection could mean in practice
Imon Banerjee, Ph.D., the Mayo Clinic AI researcher who led the study, framed the goal in the news release. "We wanted to determine whether AI could recognize subtle patterns that are imperceptible to the human eye in routinely acquired B-mode ultrasound videos and identify patients with LVOT obstruction earlier, enabling timely confirmatory Doppler evaluation and referral when appropriate."
He was clear about the model's place in care. "This technology is intended to complement, not replace, Doppler echocardiography. By enabling earlier identification of patients with potential LVOT obstruction, it could escalate timely detection and prompt confirmatory Doppler measurements, stress testing, or referral to an HCM specialty center."
The practical implication is a wider safety net. A person who has a routine echo for shortness of breath, a heart murmur or a pre-surgery workup could have the same clip run through a screening tool, even at a clinic without cardiology specialists on site. If it flags a possible obstruction, the patient can be sent for the specialized testing that would otherwise never have been ordered.
The work is still early. The external validation involved only 46 patients, and the authors describe next steps that include broader prospective testing across more platforms and populations. Anyone with symptoms such as unexplained fainting, chest pain during exertion or a family history of sudden cardiac death should raise them with a doctor rather than wait for a screening tool to become widely available.
Also read: After 60, omega-3 may help your body hold onto muscle even when you slow down
Also read: The way you sleep could be quietly affecting your back and your heart
This article is made and published by Jesper Bengtson, who may have used AI in the preparation.