Mammogram model identified heart and blood vessel disease in women
Detection scores were 0.79 for high blood pressure, 0.78 for heart disease from reduced blood supply and 0.86 for stroke.
Limit: The study assessed existing disease, not whether using the model improved health
Edited by Boris Djordjevic · AgingResearch.News
Researchers studied 29,921 women with a median age of 54 who underwent mammography at a specialist referral center. The retrospective cohort study analyzed 97,364 examinations performed between 2011 and 2025. Of the women, 5,485 had breast cancer.
The team developed a deep learning model, a type of artificial intelligence, to identify existing cardiovascular conditions from the images. Medical records and supporting clinical information supplied the diagnoses used for comparison. The model’s area under the curve scores measured its ability to distinguish women with each condition from those without it. It scored 0.79 for high blood pressure, 0.78 for heart disease caused by reduced blood supply and 0.86 for stroke.
Why it matters
Cardiovascular disease threatens health in later life. The authors suggested that adding cardiovascular assessment to existing breast-screening workflows may help reduce underdiagnosis in women.
Caveats
This retrospective study assessed existing disease, not whether using the model improved health. It used data from one center, excluded women without available clinical data and was reported as a conference abstract.
The paper
Detecting cardiovascular diseases from mammography using deep learning
