Overview

• Researchers from The George Institute, UNSW and the University of Sydney reported the results in Heart, the journal of the British Cardiovascular Society.
• The deep learning model was trained and validated on 49,196 Lifepool participants in Victoria with a median follow-up of about nine years, during which roughly 3,392 first major cardiovascular events occurred.
• Using only mammographic features plus age, the algorithm predicted 10-year risk as accurately as established calculators that rely on clinical data such as blood pressure and cholesterol.
• Because it requires no blood tests or detailed medical history, the approach could provide a cost-effective cardiovascular risk check layered onto routine breast screening.
• The authors emphasize the need for external validation in diverse populations and assessment of implementation and equity considerations before any clinical rollout.