AI-Enabled Focused Cardiac Ultrasound: Expanding Screening Access for Moderate to Severe AS

Article Impact Level: HIGH
Data Quality: STRONG
Summary of  JAMA Cardiology. https://doi.org/10.1001/jamacardio.2026.3829 
Dr. Eunjung Lee et al.

Points

Mayo Clinic researchers evaluated an artificial intelligence protocol enabling personnel without prior ultrasound experience to acquire and interpret diagnostic focused cardiac ultrasound examinations.
Aortic stenosis affects approximately seven percent of adults aged seventy-five and older, remaining the leading indication for surgical or transcatheter heart valve interventions globally.
Nine novice operators completed four hours of standardized training before capturing focused cardiac ultrasound images utilizing automated deep learning guidance and interpretation algorithms.
Diagnostic validation demonstrated successful processing in ninety-seven percent of examinations, accurately identifying ninety-three percent of moderate or severe cases and achieving ninety-six percent specificity.
Approximately ten percent of imaging studies required expert specialist review, establishing a scalable screening model to identify high-risk patients needing comprehensive echocardiography confirmation.

Summary

This study evaluated the diagnostic performance and feasibility of utilizing artificial intelligence (AI)-guided focused cardiac ultrasound (FoCUS) acquired by novice operators to screen for moderate or greater aortic stenosis (AS). Aortic stenosis affects approximately 7% of individuals aged 75 and older and represents the primary indication for valve interventions globally. Led by Gal Tsaban and Jared Bird at Mayo Clinic, presented at the 2026 ESC Congress and published in JAMA Cardiology, the investigation evaluated whether deep learning algorithms could democratize point-of-care AS screening in resource-constrained healthcare environments.

To evaluate prospective clinical performance, nine research staff members without prior clinical or sonographic experience underwent four hours of targeted training. Utilizing AI guidance for image acquisition, these novice operators performed handheld FoCUS examinations across cohort participants. The deep learning algorithm successfully analyzed nearly 97% of the acquired exams. Quantitative diagnostic evaluations demonstrated a 93% sensitivity for identifying moderate or more severe AS and a 96% specificity for ruling out disease, while approximately 10% of examinations were flagged for secondary specialist review.

Integrating AI-guided acquisition with expert overread mitigated false-positive findings, though some true positive cases were omitted upon secondary evaluation. The authors conclude that AI-enabled FoCUS acquired by minimally trained operators represents a highly scalable screening strategy to expand access to early AS detection. However, this screening approach is not intended to replace comprehensive echocardiography, and patients identified with suspected moderate or greater AS still require formal diagnostic imaging to confirm severity prior to therapeutic intervention.

Link to the article: https://jamanetwork.com/journals/jamacardiology/article-abstract/2853493

References

Lee, E., Naser, J. A., Kane, C. J., Kovac, J. L., Greason, C., Killalea, M. M., Gulati, M. A., Jackson, J. I., Borgeson, J., Schonfeld, D. A., Malins, J. G., Anisuzzaman, D. M., Crestanello, M. M., Zacher, J., Camalan, S., Thaden, J. J., Anand, V., Nkomo, V. T., Padang, R., … Tsaban, G. (2026). Artificial intelligence–enabled acquisition and interpretation for screening aortic stenosis. JAMA Cardiology. https://doi.org/10.1001/jamacardio.2026.3829