AI Trained on Mammograms Can Detect Heart Disease in Women

A new study from Sheba Medical Center in Israel has demonstrated that artificial intelligence can identify signs of heart disease in mammography images, which are typically used to screen for breast cancer. The AI model was trained by researchers, led by Dr. Vianna Kopelend, to detect indicators of cardiovascular disease, a leading cause of death in women that is often undiagnosed.

The research, presented at the European Society of Cardiology annual congress in Munich, analyzed approximately 100,000 mammograms from 30,000 women. The AI achieved an 86% accuracy rate in identifying women who had suffered a stroke, 79% for hypertension, and 78% for coronary artery disease, based solely on the mammograms. These results were consistent across age groups and unaffected by a cancer diagnosis.

Dr. Kopelend highlighted that this AI application leverages an existing, common screening tool to provide additional diagnostic insights without requiring further tests. She noted that many women's heart disease symptoms are overlooked because they differ from men's. Currently, the model identifies existing conditions, but the goal is to expand its capability to detect future risk, such as in women who experienced pre-eclampsia during pregnancy.

The research team is working to enhance the AI's accuracy and broaden the range of heart conditions it can identify. The practical advantage lies in mammography's widespread availability, relatively low cost, and minimal radiation exposure, making it accessible even in resource-limited healthcare settings. This development could lead to earlier diagnoses for many women.

Sheba Medical Center is incorporating this AI platform into its broader smart hospital initiative. The development is also moving towards commercialization, with plans to establish a company and secure funding. The findings will also be presented at Sheba's ARC Summit innovation conference in October.