Mammogram AI Found Prior Stroke at 86%, but Clinical Use Still Needs Validation
The study points to a possible way to extract cardiovascular signals from breast scans already taken in routine care. Its reported results are promising, but the path to clinical use still runs through accuracy, reliability and false-result reduction.
Listen to this story
The audio brief
Story brief
3 key pointsA machine-learning system developed by researchers in Israel found cardiovascular signals in mammograms, with reported reliability of 86% for identifying prior stroke, 79% for high blood pressure, and 78% for coronary heart disease. The analysis covered 97,364 scans from 29,921 women and used linked medical records, suggesting an existing breast-screening workflow could eventually flag heart risks without new...
- 01
The dataset included 97,364 mammograms from 29,921 Israeli women averaging 54 years old.
- 02
Stroke was the strongest result at 86%; high blood pressure reached 79%, and coronary heart disease 78%.
- 03
Results were reported as consistent across age groups and regardless of whether patients also had cancer.
Mammography is designed to find breast cancer, but researchers in Israel say its images also hold signals associated with cardiovascular conditions. Their machine-learning model identified women who had suffered a stroke from mammograms alone with 86% reported reliability. The work points to a possible second use for a familiar screening exam, while experts say accuracy and reliability still must be established before clinical implementation.
Researchers analyzed 97,364 mammogram scans from 29,921 women in Israel, whose average age was 54, and cross-referenced the scans with medical records. They trained the model to distinguish women with documented cardiovascular conditions from those without them. In the study group, 16% had high blood pressure, while 2.5% had coronary heart disease and 2.5% had experienced a stroke.
What the model picked up
The model’s reported performance varied by condition. Stroke was its strongest result, followed by high blood pressure and coronary heart disease. The results were reported as consistent across age groups and whether or not a woman also had cancer.
The model identified women who had suffered a stroke from mammograms alone with 86% reported reliability.
The reported reliability for distinguishing women with high blood pressure from those without it was 79%.
The reported reliability for distinguishing women with coronary heart disease from those without it was 78%.
The figures are described as reliability. The study account does not provide a further definition of the underlying measurement.
A second use for an existing scan
Dr. Viana Copeland of Tel Aviv University presented the findings at the European Society of Cardiology’s annual congress in Munich on August 27. Because mammography is already widely used, Copeland said analyzing breast scans for heart-health information could offer a scalable approach without an additional imaging examination.
Experts said that, if further proven, the approach could make breast screening dual-purpose: detecting breast cancer while also flagging cardiovascular concerns. Copeland said mammography reaches many women in midlife, an important period for recognizing cardiovascular risk.
The remaining clinical gap
Researchers are working to improve the model’s accuracy, reduce false results and expand the range of heart conditions it can detect. Experts called the findings compelling, but said the challenge is to establish accuracy and reliability before the approach moves into clinical care.
Sources
- theguardian.comAI can detect heart disease in women using mammograms, study suggests
