A deep learning model developed by researchers at Emory Healthcare and Mayo Clinic may help assess cardiovascular risk in women by analyzing calcification seen in routine mammograms, according to findings being presented at the American College of Cardiology’s Annual Scientific Session.
The study focused on breast arterial calcification; calcium deposits visible on mammograms that are not typically quantified or reported in standard screenings. Researchers used AI to automatically segment and measure the extent of calcification, then linked these findings with cardiovascular outcomes recorded in electronic health records from more than 56,000 patients.
“Our study showed that breast arterial calcification is a good predictor for cardiovascular disease, especially in patients younger than age 60,” said Dr. Theo Dapamede, a postdoctoral fellow at Emory University and lead author of the study. “If we are able to screen and identify these patients early, we can refer them to a cardiologist for further risk assessment.”
The model classified cardiovascular risk levels as low, moderate or severe based on calcification measurements, and tracked rates of death and major cardiovascular events — including heart attack, stroke and heart failure — over two- and five-year periods. Women with the highest levels of calcification (above 40 mm²) had significantly worse outcomes, with a five-year event-free survival rate of 86.4%, compared to 95.3% among those with minimal calcification (below 10 mm²).
The predictive value of calcification was strongest in women under age 80, with no significant association found in patients older than 80. The researchers emphasized that early detection may be especially valuable in younger women, who often remain underdiagnosed and undertreated for cardiovascular disease.
The AI tool is not yet available for clinical use. Further validation and regulatory approval are required before it could be integrated into routine mammography workflows. Researchers also plan to explore whether similar models could help detect risks for other conditions, such as peripheral artery disease or kidney disease.
