AI Algorithm Uses Mammograms to Accurately Predict Cardiovascular Risk in Women

Cardiovascular disease remains the leading cause of death in women worldwide, responsible for about nine million deaths annually. Despite this burden, symptoms and risk factors are often under-recognized in women, leading to fewer diagnostic tests and delayed treatment compared to men. Many women are also not offered cardiovascular risk screening in the community, making early detection especially difficult. Researchers have now developed a new machine learning model that can successfully predict heart disease risk in women by analyzing mammograms.

Researchers at the George Institute for Global Health (Sydney, Australia), in collaboration with the University of New South Wales (Sydney, Australia) and collaborators, have developed a deep learning model that analyzes mammograms to predict cardiovascular risk. Unlike traditional methods requiring multiple clinical data points, this model relies only on mammographic features and age. By integrating cardiovascular screening with breast screening, it enables a “two-for-one” approach that is less resource-intensive yet highly effective.

The model was designed and validated using mammograms from over 49,000 women across metropolitan and rural Australia, linked to hospital and death records. It was then compared to traditional risk calculators that depend on blood pressure, cholesterol, and other risk factors. The findings, published in Heart, showed that the model achieved comparable accuracy without requiring additional medical history or data.

This innovation addresses limitations of previous approaches, such as those focusing solely on breast arterial calcification (BAC), which is less reliable in older women. By using a wider range of mammographic features, the model offers stronger predictive power across diverse populations. Importantly, it could expand access to risk screening in rural and underserved areas, where mobile mammography units already reach many women free of charge.

Researchers now plan to test the model in broader and more diverse populations to validate its global applicability. If implemented widely, the approach could help reduce gender disparities in cardiovascular screening, offering women more equitable access to early detection tools. The long-term goal is to integrate cardiovascular risk prediction seamlessly into existing mammography programs worldwide.

“It’s a common misconception that cardiovascular disease predominantly affects men, resulting in underdiagnosis and undertreatment of women,” said Associate Professor Clare Arnott, Global Director of the Cardiovascular Program at The George Institute for Global Health. “By integrating CV risk screening with breast screening through the use of mammograms - something many women already engage with at a stage in life when their cardiovascular risk increases - we can identify and potentially prevent two major causes of illness and death at the same time.”

Related Links:
The George Institute for Global Health
University of New South Wales