AI-Guided Ultrasound Enables Nonexpert Screening for Aortic Stenosis

Comprehensive echocardiography is the diagnostic standard, but it requires specialized equipment and expert operators, limiting access for broad screening. Missed or delayed detection can impede timely monitoring and intervention for at‑risk patients. To help address this challenge, Mayo Clinic researchers have developed an AI-assisted approach that enables novice users to acquire focused cardiac ultrasound images and flag patients who may have clinically significant disease.

The AI-assisted workflow from Mayo Clinic combines real‑time guidance for image acquisition with a deep learning algorithm for interpretation. The algorithm was developed and validated using echocardiograms from patients across Mayo Clinic sites in Arizona, Florida, the Mayo Clinic Health System, and Rochester. Its performance was also evaluated on hand‑held ultrasound images collected by experienced sonographers to assess robustness across acquisition methods.

In a prospective study, nine research staff members without clinical or ultrasound experience received four hours of training before performing focused cardiac ultrasound examinations with AI guidance. The AI model was able to analyze nearly 97% of exams. Using this workflow, the team correctly identified 93% of patients with moderate or more severe aortic stenosis and correctly ruled out 96% of patients who did not have the condition.

About 10% of exams were flagged for review by a heart imaging specialist. The combined approach reduced false positives, although some patients with aortic stenosis were not identified. The workflow is intended for screening and triage and is not a replacement for a comprehensive echocardiogram or a physician’s evaluation. Patients flagged as potentially having moderate or more severe disease still require comprehensive echocardiography to confirm the diagnosis and define severity.

The condition affects approximately 7% of people age 75 and older and is a leading reason for valve intervention, underscoring the need for scalable screening options. Findings were published in JAMA Cardiology on August 28, 2026, and presented at the 2026 ESC Congress. 

Related Links
Mayo Clinic