‘Two-for-one’ screening uses mammograms to predict heart disease in women

Cropped shot of a woman wearing a white T-shirt. She is holding a paper made...

A new machine learning model developed by The George Institute for Global Health can successfully predict heart disease risk in women by analysing mammograms. The findings were published in Heart,¹ the official journal of the British Cardiovascular Society. Developed in collaboration with the University of New South Wales and University of Sydney, this is the first deep learning algorithm based on only mammographic features and age to predict major cardiac events with comparable accuracy to traditional cardiovascular risk calculators.

By integrating CV risk screening with breast screening through the use of mammograms, we can identify and potentially prevent two major causes of illness and death at the same time
Clare Arnott

Associate Professor Clare Arnott, Global Director of the Cardiovascular Program at The George Institute said that new ways to identify women at risk of cardiovascular disease (CVD) were needed, given that many women are not accessing or being offered CV risk screening in the community. “It’s a common misconception that CVD predominantly affects men, resulting in underdiagnosis and undertreatment of the condition in women. 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.”

The model was designed and validated using routine mammograms from over 49,000 women in metropolitan and rural areas of Victoria, Australia, linked to individual hospital and death records. Researchers then compared the model to traditional models that require multiple data points based on known CV risk factors such as blood pressure and cholesterol. “We found that our model performed just as well without the need for extensive clinical and medical data,” said A/Prof Arnott.

Previous research to date has focused on certain mammographic features such as breast arterial calcification (BAC), which has been found to be associated with cardiovascular risk in some populations. Relying on BAC alone, however, has limitations.² For example, BAC is less accurate at predicting CVD risk in older women.² “Our model is the first to use a range of features from mammographic images combined simply with age - a key advantage of this approach being that it doesn’t require additional history taking or medical record data, making it less resource intensive to implement, but still highly accurate,” Professor Arnott added.

Globally, cardiovascular disease is the leading cause of mortality in women, amounting to around 9 million deaths annually, or approximately one third of all deaths in women.³ Despite the high burden of disease, multiple studies internationally have shown that cardiovascular disease symptoms and risk factors are under-considered in women, leading to fewer diagnostic tests, specialist referrals and prescriptions in women compared to men.⁴ Conversely, mammography-based screening programmes have engaged women very effectively in some countries, with more than 67% of women in the United States and the United Kingdom participating in screening mammography.⁵^,⁶

Dr Jennifer Barraclough, Research Fellow at The George Institute, said that leveraging an existing risk screening process already widely utilised by women, means this model could serve as a cardiovascular risk prediction tool for women in diverse communities across Australia and around the world. “We hope this technology will one day provide greater, and more equitable access to screening in rural areas, as many women already benefit from mobile mammography units free of charge,” she said. “We have shown the potential of this innovative new screening tool, so we now look forward to testing the model in additional, diverse, populations and understanding potential barriers to its implementation.”

References:

Source: George Institute for Global Health

19.09.2025