Deep Learning Algorithms use mammograms and age to predict heart disease

A new machine learning model developed by the George Institute for Global Health allows women to successfully predict heart disease risk by analyzing mammograms. The results of the survey were published today heart, Official journal of the British Cardiovascular Association.

Developed in collaboration with the University of New South Wales and the University of Sydney, it is the first deep learning algorithm based solely on mammography features and age to predict major cardiac events with accuracy comparable to traditional cardiovascular risk calculators.

Associate Professor Claire Arnott, global director of cardiovascular programs at George Institute, said that given the lack of access to or offering CV risk screening in the community, there is a need for a new way to identify women at risk for cardiovascular disease (CVD).

“It is a common misconception that CVD mainly affects men and causes underdiagnosis and shortages in women. By integrating CV risk screening and breast screening through the use of mammograms, if many women are already involved in stages, an increase in cardiovascular risk will allow us to grasp two major causes and causes of death.

This model was designed and validated using routine mammograms from over 49,000 women in metropolitan and rural areas of Victoria, Australia, in relation to individual hospitals and death records. The researchers then compared the models with traditional models that require multiple data points based on known CV risk factors such as blood pressure and cholesterol.

“We found that our model works similarly without the need for extensive clinical and medical data,” A/Prof Arnott said.

Previous studies to date have focused on specific mammographic features such as breast arterial calcification (BAC), which has been found to be associated with cardiovascular risk in some populations. However, there are limitations to relying solely on BAC. For example, BAC is not very accurate in predicting the risk of CVD in older women.

Our model is the first to simply use the various features of mammographic images combined with age. An important advantage of this approach is that it does not require additional history or medical record data and requires less resource implementation, but is still very accurate. ”

Claire Arnott, Global Director of Cardiovascular Programs, George Institute

Globally, cardiovascular disease is the main cause of deaths in women, representing approximately 9 million deaths per year, or about a third of all deaths in women. Despite the high burden of disease, several studies have shown that cardiovascular disease symptoms and risk factors are underestimated in women, with fewer diagnostic tests, expert referrals and prescriptions in women compared to men.

Conversely, mammography-based screening programs have very effective involvement of women in some countries, with over 67% of women in the US and UK taking part in screening mammography.

Dr. Jennifer Balaclau, a researcher at the George Institute, said that leveraging existing risk screening processes already widely used by women means that the model could serve as a cardiovascular risk prediction tool for women in diverse communities across Australia and around the world.

“We hope that one day we will provide more, more equitable access, as many women will already benefit from the mobile mammography unit for free,” she said.

“We are looking forward to testing the model in an additional diverse population and understanding the potential barriers to its implementation, as we demonstrate the potential of this innovative new screening tool.”