AI analysis of routine mammograms can predict heart attack and stroke risk in women

Nearly 70% of American women aged 45 and older are current with their mammography screening. Fewer than 40% report knowing their cholesterol levels. That mismatch may contain a public health opportunity.

A study published March 9, 2026, in the European Heart Journal demonstrates that artificial intelligence can analyze standard mammogram images to detect calcification in breast arteries -- and that the degree of calcification independently predicts a woman's risk of heart attack, stroke, heart failure, and cardiovascular death. The research, led by Dr. Hari Trivedi at Emory University, included 123,762 women with no known cardiovascular disease.

What the AI measures

Breast arterial calcification (BAC) refers to calcium deposits that build up in the walls of arteries running through breast tissue. These deposits are visible on standard mammography X-rays and are already routinely noticed by radiologists, though they are typically noted and then ignored. The calcium signals arterial hardening -- the same process that drives cardiovascular disease elsewhere in the body.

The AI system categorized women's breast arterial calcification as absent, mild, moderate, or severe. The researchers then tracked whether these women went on to develop major cardiovascular events.

The risk gradient

The results showed a clear dose-response relationship. Women with mild calcification were approximately 30% more likely to experience serious cardiovascular disease compared to women with no calcification. Moderate calcification raised the risk by more than 70%. Severe calcification doubled to tripled the risk.

These associations held after adjusting for traditional cardiovascular risk factors including diabetes and smoking. They also held in women under 50 -- a group typically considered low-risk for heart disease. The study spanned multiple races and two major U.S. health systems, giving the findings broader applicability than single-center studies.

"We found that the more calcium visible in the breast arteries on a mammogram, the higher a woman's risk of a serious heart event," Trivedi said. "This was true even in younger women under 50."

Two screenings from one scan

The practical appeal is straightforward. Women already attend mammography appointments for breast cancer screening. Adding cardiovascular risk assessment to the same image requires no additional scanning, no extra appointment, and no added cost to the patient. The AI tool runs on the existing image.

Heart disease is the leading cause of death in women worldwide, yet women are consistently underdiagnosed and undertreated compared to men. Part of the problem is that traditional cardiovascular risk assessment depends on tests -- cholesterol panels, blood pressure monitoring, risk calculators -- that many women never receive. Mammography reaches a far larger population.

In an accompanying editorial, Professor Lori B. Daniels of UC San Diego noted the scale of the opportunity. Two-thirds of women aged 50 to 69 in the European Union reported having a mammogram within the prior two years. Integrating cardiovascular screening into that existing platform could reach tens of millions of women annually.

From observation to action

Breast arterial calcification has been observed on mammograms for decades. What has been missing is standardization -- a consistent way to quantify it -- and a clear pathway from detection to clinical action. The AI tool provides the measurement. What happens next is a clinical workflow question.

Trivedi envisions a system where a positive BAC finding prompts a conversation with the patient's primary care physician about cardiovascular prevention: cholesterol testing, blood pressure management, lifestyle modification, or medication. The team is now planning a clinical trial to test these implementation steps.

The study's limitations include its observational design -- it demonstrates association and prediction, not causation. Whether acting on BAC findings actually reduces cardiovascular events requires the prospective clinical trial the team has proposed. The AI tool's performance also needs validation across different mammography equipment and imaging protocols before widespread deployment.

Daniels concluded that the evidence now supports moving breast arterial calcification from incidental observation to implemented screening tool, "leveraging a touchpoint women already trust, to advance prevention for what remains the leading cause of death among women."