AI uses mammograms to accurately predict women’s heart disease risk

AI uses mammograms to accurately predict women’s heart disease risk | Image Credit: © Valerii Apetroaiei - © Valerii Apetroaiei - stock.adobe.com.

Investigators have published data in the journal Heart indicating efficacy for an AI algorithm using only routine mammogram images and age to assess major cardiovascular disease risk in women.¹

Equal efficacy to modern risk scores using age and clinical factors was highlighted for the algorithm. Based on this data, investigators suggested cardiovascular risk assessment through routine mammograms may be an effective, cost-effective solution by relying on existing health infrastructure.¹

“In contrast with what is commonly thought, breast cancer causes only about 10% of the total deaths globally compared with those resulting from cardiovascular disease,” wrote investigators. “Mammography may therefore represent a 'touch point' for raising awareness about cardiovascular risk and disease in women.”¹

Study details and participant profile

The trial was conducted to evaluate the accuracy of routine mammogram images undergoing an automated AI analysis of the full range of internal breast structure and characteristics to determine cardiovascular risk in women. Data from 49,196 women aged a mean of 59 years was obtained from the Lifepool cohort registry.¹

Baseline information obtained included age, alcohol intake, smoking status, weight, and histories of diabetes, high blood pressure, or cholesterol drug use, and blood thinner use. Additionally, menopause status, hormone therapy use, reproductive history, and factors influencing internal breast structure, such as radiation, surgery, and cancer, were reported.¹

Currently, smoking was reported in 5% of participants, having a body mass index over 25 in 62%, and type 2 diabetes in 6%. Thirty-three percent took drugs for high cholesterol, 27% for high blood pressure, and 11% a blood thinner.¹

AI algorithm matches leading risk models

Of 49,196 participants, 3392 experienced a first cardiovascular event during the follow-up period. This included coronary artery disease in 2383, heart attack in 656, stroke in 434, and heart failure in 731.¹

In these patients, the AI algorithm using mammography features and patient age had a concordance index of 0.72.² Investigators noted this performance is equally effective to current prediction models such as the New Zealand ‘PREDICT’ tool and the American Heart Association ‘PREVENT’ equations.

According to investigators, women are often underscreened and undertreated for cardiovascular disease and its associated risk factors, with risk prediction algorithms often underperforming.¹ While arterial calcium deposits (BAC) and breast tissue density are used in current risk assessment models, BAC is not linked to obesity.

Additionally, a negative association has been reported between BAC and smoking. This suggests it is not effective for cardiovascular risk prediction by itself.¹

“Risk assessments based on mammography may be a novel opportunity for improving cardiovascular risk screening in women,” concluded investigators.²

Importance of nontraditional risk factors

This screening method may support women with nontraditional risk factors of heart attack, which are more common in this population.³ The increased presence of nontraditional risk factors has been confirmed in a trial from Mayo Clinic researchers, including women aged under 65 years.

In this population, over 50% of heart attacks were linked to nontraditional risk factors. Additionally, women were more likely to have their primary cause of heart attack misdiagnosed compared to men.³

Misdiagnoses were more common in cases of spontaneous coronary artery dissection in younger, otherwise healthy women. Investigators concluded sex-specific considerations are vital for preventing cardiovascular events.³

References

1. AI algorithm based on routine mammogram + age can predict women’s major cardiovascular disease risk. BMJ Group. September 16, 2025. Accessed September 23, 2025. https://www.eurekalert.org/news-releases/1098150.
2. Barraclough JY, Gandomkar Z, Fletcher RA, et al. Predicting cardiovascular events from routine mammograms using machine learning. Heart. 2025. doi:10.1136/heartjnl-2025-325705
3. Krewson C. Nontraditional causes linked to many heart attacks in younger women. Contemporary OB/GYN. September 17, 2025. Accessed September 23, 2025. https://www.contemporaryobgyn.net/view/nontraditional-causes-linked-to-many-heart-attacks-in-younger-women.