Aug. 16, 2025
Mayo Clinic cardiovascular researchers created a solution for detecting cardiac amyloidosis using artificial intelligence (AI) and echocardiography. The findings of the multicenter collaboration were published in the European Heart Journal.
The AI-enhanced echocardiography model developed by Mayo Clinic and an AI echocardiography company is the initial and only tool of its kind. "Our model was approved by the FDA as a breakthrough device. It became the first commercially available AI echocardiography device to screen for amyloid cardiomyopathy," says Patricia A. Pellikka, M.D., a cardiologist, echocardiographer and immediate past chair of Echocardiography and Cardiovascular Imaging at Mayo Clinic in Rochester, Minnesota. Dr. Pellikka was senior author of the study.
With cardiac amyloidosis, an abnormal protein, called an amyloid, builds up in the heart, causing it to stiffen. This rare and progressive type of heart failure is often missed because the symptoms and imaging features can look like other heart conditions. "Amyloid cardiomyopathy is associated with thick walls of the left ventricle and other cardiac chambers," says Dr. Pellikka. "Some conditions associated with increased thickening of the left ventricle walls include hypertensive heart disease, hypertrophic cardiomyopathy and aortic stenosis. They can coexist with amyloid cardiomyopathy and may present with symptoms including fatigue, shortness of breath and ankle swelling."
Inventing better detection
This research stems from Dr. Pellikka's earlier work with Mayo Clinic colleagues of developing an AI-electrocardiogram model to detect left ventricular systolic dysfunction from a 12-lead ECG. And that led to exploring what might be possible by applying deep learning to echocardiography. She helped build a Food and Drug Administration (FDA)-cleared model to automatically interpret an echocardiogram to detect heart failure with preserved ejection fraction (HFpEF). HFpEF, a common type of heart failure, is associated with high morbidity and mortality and can be difficult to diagnose. About 15% of patients with HFpEF have cardiac amyloidosis.
"Amyloid cardiomyopathy, which can be a type of HFpEF, was the next appropriate project to pursue, given its difficulty in detection, the important role of echocardiography in screening and the recent availability of impactful new treatments for it," says Dr. Pellikka.
Superior screening
The study findings, published in the European Heart Journal show that the AI-enhanced echocardiography model was highly accurate. "When we tested the model in an independent, multinational cohort with a 22% prevalence of cardiac amyloidosis, it performed well for both discrimination and classification: area under the receiver operating characteristic curve (AUROC) 0.93, sensitivity 85%, specificity 93%, positive predictive value 78%, negative predictive value 96%," says Dr. Pellikka. "The AI echocardiography model, based on nothing more than a single standard echocardiographic video clip, was superior to previously validated scores that we have been using to screen for amyloid cardiomyopathy. These scores combine clinical variables and echocardiographic measurements."
What the researchers didn't expect to find, while exploring the database for building the model, was that 0.53% of patients receiving an echocardiogram have features that suggest amyloid cardiomyopathy. "But the diagnosis is not explored with confirmatory testing. Better screening is clearly needed," says Dr. Pellikka.
Next steps
This AI-enhanced echocardiography model is a vital tool that can help identify patients promptly. New treatments for amyloidosis are most effective when administered early.
The technology is being used at multiple centers in the U.S. "The AI echocardiography model needs to be tested prospectively in clinical practice to fully understand which patients should receive screening with it and the limitations," says Dr. Pellikka. "The AI echocardiography model could be implemented in any echocardiography lab to improve accuracy of screening for cardiac amyloid."
For more information
Slivnick JA, et al. Cardiac amyloidosis detection from a single echocardiographic video clip: A novel artificial intelligence-based screening tool. European Heart Journal. In press.
Refer a patient to Mayo Clinic.