AI in Echocardiography Shows Promise for Earlier Detection of Cardiac Amyloidosis

Artificial intelligence is beginning to reshape the way cardiologists approach one of the most underdiagnosed contributors to heart failure: cardiac amyloidosis. New findings from Ultromics, presented at the American Society of Echocardiography’s 2025 Scientific Sessions in Nashville and published as an abstract in the Journal of the American Society of Echocardiography, suggest that AI-powered echocardiography can help physicians identify the disease earlier and with greater accuracy.

Cardiac amyloidosis has moved from a rare, late-stage discovery to a condition now at the center of cardiology practice. The arrival of targeted therapies such as tafamidis and acoramidis—shown to slow disease progression and improve survival—has intensified the push for earlier detection. Yet in day-to-day practice, up to two-thirds of cases remain undiagnosed, often until symptoms are advanced and treatment options are less effective. Ultromics’ study points to a potential solution: using AI to detect subtle, early signs of amyloid infiltration on standard heart ultrasounds.

The study evaluated more than 4,800 patient cases across 17 hospitals in the U.S. and U.K., modeling how Ultromics’ EchoGo® Amyloidosis tool could alter referral decisions. In traditional practice, cardiologists often rely on wall thickness as a marker for suspicion. This method identified about 65% of patients in low-prevalence settings. When the AI model was applied, correct referral rates improved to as high as 80%, enabling more patients to be flagged sooner while avoiding a portion of unnecessary testing.

In higher-prevalence settings, the AI proved equally valuable, reducing unnecessary referrals by nearly 18% while maintaining strong detection rates. The results were consistent across sites in both countries, highlighting the technology’s potential scalability and reliability in routine clinical environments.

“Too often, patients with cardiac amyloidosis are diagnosed only after years of unexplained symptoms and irreversible damage,” noted Dr. Ashley Akerman, Director of Clinical Sciences at Ultromics and lead author of the study. “Our findings suggest that integrating EchoGo® Amyloidosis into routine echocardiography could better identify at-risk patients, streamline referrals, and ensure timely access to therapies that can change the trajectory of the disease.”

EchoGo® Amyloidosis works by analyzing echocardiograms at the pixel level, capturing patterns too subtle for the human eye. The model has been trained and validated on over 9,700 echocardiographic videos from more than 7,000 patients across 15 international centers, then tested on an additional 2,700 patients spanning 18 sites. In this broad, multi-ethnic dataset, it achieved an area under the curve (AUC) of 0.93—a measure of accuracy considered strong for clinical use.

The study adds to a growing body of evidence supporting Ultromics’ EchoGo® platform, which is FDA-cleared, Medicare-reimbursed, and already deployed at leading academic centers such as UChicago Medicine, Northwestern, and City of Hope. Beyond amyloidosis, the platform is being used to support earlier detection of a range of cardiovascular conditions, providing clinicians with consistent, automated insights that can complement expert interpretation.

For cardiologists, the message is clear: AI tools like EchoGo® Amyloidosis could play a key role in closing the diagnostic gap for a disease where timing makes all the difference.