Innovative AI Enhances Women's Heart Risk Assessment Using Routine Mammograms

A groundbreaking AI-based approach has been developed to assess women's risk of major cardiovascular events through routine mammogram images. Published in the journal Heart, this innovative method utilizes only standard mammogram scans and age data to predict a woman's likelihood of experiencing conditions such as coronary artery disease, heart attack, stroke, or heart failure over a ten-year period. The study involved analyzing data from over 49,000 women in Australia, with follow-up results revealing that the AI model matched or slightly exceeded traditional risk assessment tools.

This approach leverages existing mammography infrastructure, making it a potentially cost-effective 'two-for-one' screening method that could streamline cardiovascular risk detection in women. Traditional risk calculators often underperform in women due to complex clinical data requirements and a lack of awareness regarding heart disease in women. In contrast, this AI model examines internal breast structures, including tissue density and calcium deposits, which have been linked to cardiovascular health, although previous methods relied on isolated markers like breast arterial calcium deposits (BAC) that were limited.

The AI algorithm considers a comprehensive range of breast features alongside demographic and health information such as menopausal status, reproductive history, and medication use. During nearly nine years of follow-up, data showed that women who experienced a cardiovascular event could be accurately identified based on their mammogram features combined with age. The model's performance was comparable to or better than existing risk scores like New Zealand's PREDICT or the American Heart Association's PREVENT calculator, without needing extensive clinical data.

This discovery highlights a promising shift in preventive health, where routine mammography could serve a dual purpose—early detection of breast health issues and prediction of cardiovascular risk. Experts emphasize that as cardiovascular disease remains a leading cause of death worldwide, especially among women, using mammography as a risk assessment tool could enhance awareness and early intervention.

However, the researchers acknowledge limitations such as differences in imaging equipment and the reliance on self-reported health data. Despite these challenges, the authors believe this approach offers a practical, accessible option for early cardiovascular risk screening, particularly because it does not require additional testing or data collection. As machine learning models continue to evolve, integrating such technology could significantly improve health outcomes for women, leveraging existing healthcare visits to identify those at high risk before clinical symptoms appear.

In an accompanying editorial, experts from the University of Sydney note that traditional risk scores are less effective in women and underscore the importance of increasing awareness about heart disease risk in women. They suggest mammography could serve as a critical touchpoint for education and prevention efforts, although they also recognize the challenges of integrating new tools into routine practice.

This innovative research underscores the potential for artificial intelligence to revolutionize preventive medicine, providing more accurate and accessible risk assessments using routine imaging techniques.