The model addresses misconceptions that cardiovascular disease primarily affects men, highlighting the underdiagnosis and undertreatment of women, and emphasizes the importance of integrated screening to prevent major causes of death among women.

A new study published in the journal Heart reveals that an AI algorithm analyzing routine mammogram images, combined with age, can predict a woman's 10-year risk of major cardiovascular disease as effectively as traditional risk assessment tools.

This AI model was trained and validated on over 49,000 mammograms from women in Victoria, Australia, and demonstrated strong performance without needing additional clinical data like blood pressure or cholesterol levels.

The research involved nearly 50,000 women with an average age of 59, tracking their health over almost nine years and recording over 3,300 cardiovascular events such as heart attacks, strokes, and heart failure.

Unlike previous methods that relied solely on mammographic features like breast arterial calcification, this new approach combines various features with age for improved accuracy in predicting cardiovascular risk.

Utilizing routine mammograms leverages existing screening programs, especially in rural areas where mobile units are already in operation, making the approach resource-efficient and scalable.

An editorial accompanying the study underscores the underrecognition of cardiovascular disease in women, noting that breast cancer accounts for only about 10% of deaths globally compared to cardiovascular disease, and suggests mammography could help raise awareness.

Further validation across diverse populations and ethnicities is ongoing to ensure the model's generalizability and facilitate its integration into routine clinical practice.

Despite its promise, implementing this AI-based risk assessment faces challenges, including integration into current clinical workflows and healthcare systems.

Experts believe that incorporating cardiovascular risk assessment into existing mammography programs could improve early detection and health equity, particularly in underserved and rural communities.

Limitations of the study include variability in imaging data across different scanners, reliance on self-reported clinical data, and the dependency of deep learning models on their training datasets.

Given that cardiovascular disease remains the leading cause of death among women globally, with around nine million deaths annually, and women often receive less diagnostic attention, this approach could significantly impact women's health.