AI spots cardiovascular concerns beyond BACs on mammograms

Artificial intelligence-powered mammograms may offer a way to identify women at increased risk of common cardiovascular diseases without the need for additional imaging. 

Breast arterial calcifications (BACs) are the most common finding believed to be associated with cardiovascular diseases on breast imaging, but experts have developed new technology that can look beyond BAC to predict patients’ CVD risk. New research presented during the 2026 European Society of Cardiology Congress suggests that AI also can detect signs of hypertension, ischemic heart disease and stroke. 

Researchers trained a deep learning model to identify mammographic features associated with each condition using clinical and imaging data. They then evaluated its ability to distinguish women with and without the diseases. The study included nearly 30,000 women who underwent 97,000 mammograms, with a median age of 54.  

According medical histories, hypertension had been diagnosed in 16% of the women, ischemic heart disease in 2.5% and stroke in 2.5%. AI was able to reliably catch signs of these diseases on the patients’ mammograms. The model demonstrated its strongest performance for stroke, with an area under the receiver operating characteristic curve (AUROC) of 0.86. Performance also was promising for hypertension, with an AUROC of 0.79, and ischemic heart disease, with an AUROC of 0.78.