AI Reads Routine Mammograms and Spots Heart Disease in Women, Major Study Finds

Doctors have discovered a way to use routine mammograms — the X-ray scans that screen for breast cancer — to spot heart disease, the world’s leading and frequently underdiagnosed cause of death in women. The findings, presented on 27 August 2026 at the European Society of Cardiology’s annual congress in Munich, the world’s largest heart conference, showed that a machine-learning model analysing standard breast scans could reliably identify women who had suffered a stroke, had high blood pressure, or were living with coronary heart disease.

The study in numbers

The research team, led by Dr Viana Copeland of Tel Aviv University in Israel, examined 97,364 breast scans from 29,921 women with an average age of 54. By cross-referencing the women’s medical records, the doctors established the ground truth for their model: 16% of the women had high blood pressure, 2.5% had coronary heart disease, and 2.5% had experienced a stroke.

The machine-learning model was then trained to detect these conditions from mammogram images alone. The results were striking:

• Stroke: the model identified women who had suffered a stroke 86% of the time
• High blood pressure: 79% reliability at distinguishing women with the condition from those without
• Coronary heart disease: 78% reliability

Crucially, the results were consistent regardless of the women’s age, and whether or not they also had cancer. That consistency matters enormously for a screening tool intended to run silently alongside an existing national programme.

Why this matters

The clinical context is sobering. Despite being the leading cause of death in women worldwide, cardiovascular disease (CVD) is consistently underdiagnosed and undertreated in women. For decades, heart disease has carried the reputation of being a “man’s disease”, a myth that has real consequences: when women do seek medical help, their CVD is often already advanced.

“A common finding in our medical centre, and around the world, is that when women do seek medical help, their CVD is already advanced,” Dr Copeland told delegates in Munich. “On the other hand, many women do attend routine breast cancer screening, even when they haven’t sought care for cardiovascular symptoms.”

That asymmetry is the insight powering the research. Hundreds of millions of women undergo mammograms every year, and attendance at breast screening programmes tends to be far higher than attendance at dedicated cardiovascular check-ups. Because mammography is already widely deployed, analysing breast scans for heart health, as Copeland put it, “could potentially offer a scalable approach without requiring an additional imaging examination”. She added that mammography “also reaches many women in midlife, an important period for recognising and addressing cardiovascular risk”.

The science behind the signal

The underlying signal the AI exploits is breast arterial calcification (BAC) — calcium deposits visible in the arteries of the breast on standard mammography films. Radiologists have known about BAC for decades, but manually grading it scan-by-scan has never been practical at population scale, and the finding has historically been ignored in breast screening reports because it is irrelevant to cancer detection.

Machine learning changes the economics of that entirely. An automated model can quantify calcification patterns across every scan in a screening programme at negligible marginal cost, learning correlations between vascular signatures and cardiovascular outcomes that no human reader would have time to chase. Earlier work — including a 2025 study from the George Institute for Global Health published in the journal Heart, and a 2023 American Heart Association analysis — had already established that BAC predicts cardiovascular risk, with women showing severe calcification at roughly double the risk of those with none. The new Israeli study pushes the idea from “predictive association” toward “detectable diagnosed disease” using nothing but the scan a woman was already getting.

Expert reaction

Elena Arbelo, an expert member of the European Society of Cardiology’s communication committee, called the findings “compelling”. “A mammogram may one day do more than look for breast cancer – it may also offer a window on to cardiovascular health,” she said. “That matters because CVD in women is still too often recognised late.” She cautioned, however, on the road ahead: “The challenge now is to establish accuracy and reliability – to move from experimentation to clinical implementation.”

Dr Sonya Babu-Narayan, a consultant cardiologist and clinical director of the British Heart Foundation, highlighted the gender gap the technology could help close. “Despite this, the myth persists that it’s only a ‘man’s disease’, meaning that when it comes to the heart, women are disproportionally unaware, unheard, underdiagnosed, undertreated and typically underrepresented in clinical research,” she said. “It is exciting to think that if the approach used in this large study from Israel is further proven, it could lead to better and earlier cardiovascular disease detection and prevention for women. AI could one day allow breast cancer screening programmes to become dual-purpose, helping to flag women with the highest risk of dangerous cardiovascular disease, as well as spotting breast cancer early.”

What comes next

The team of doctors and medical researchers is now working to improve the accuracy of the AI model and reduce false results, while expanding the range of heart conditions it can detect. The path from congress presentation to clinical tool typically runs through prospective validation trials, regulatory clearance, and integration with national screening infrastructure — a process measured in years, not months.

But the direction of travel is clear. Breast cancer screening programmes already reach a huge proportion of women in midlife, the exact demographic where cardiovascular risk begins to accelerate and where symptoms are most often dismissed. If dual-purpose mammography survives validation, the marginal cost of a cardiovascular risk screen could approach zero — no extra radiation, no extra appointment, no extra hardware. Just an algorithm reading what the scanner already saw.