The mammogram already knows about your heart. Nobody was reading it that way
A team in Israel ran 97,364 breast scans through an AI model and found something the scans were never meant to show: which women had heart disease, high blood pressure or a past stroke.
Every year, hundreds of millions of women get a mammogram — a routine X-ray of the breast that looks for cancer. The picture is taken, the radiologist checks for tumours, and that's the end of it.
Doctors in Israel looked at the same pictures again and asked a different question. They took 97,364 scans from 29,921 women, average age 54, matched them against medical records, and trained a machine-learning model to spot something else entirely: the state of the woman's heart. It worked. The model identified women who had suffered a stroke, from the mammogram alone, 86% of the time. For high blood pressure it was 79%, for coronary heart disease 78%. The findings were presented in Munich at the European Society of Cardiology's annual congress, the world's largest heart conference.
Heart disease is the leading cause of death in women worldwide, and it is the one they are least likely to be told about in time. Dr Viana Copeland of Tel Aviv University, who presented the work, put the problem in one sentence: when women do seek help, their cardiovascular disease is already advanced.
Here is the part that makes this different from most medical AI news. Nobody has to build anything new. The scan already exists, the woman already came in, the machine already took the picture. The information about her arteries was sitting in that image the whole time — unread. Copeland's word for it was "scalable": no extra imaging examination required.
A test she already takes, for a disease she wasn't being tested for.
And the timing is the quiet detail. Mammography reaches women in midlife — the exact years when heart risk starts to build and is still cheap to fix. The results held regardless of age, and regardless of whether the woman also had cancer.
Women in their 30s and 40s who assume heart trouble is something to think about later, and who have absorbed the idea — Dr Sonya Babu-Narayan of the British Heart Foundation called it a myth that persists — that this is a man's disease. Their daughters and sons, who will be the ones sitting in the waiting room in twenty years. And anyone who has ever been sent home from a clinic with a shrug: this is a story about a symptom nobody looked for, found by accident in a picture taken for something else.
The doctors and researchers say they are working to improve the model's accuracy, cut down false results, and widen the list of heart conditions it can detect. Elena Arbelo of the European Society of Cardiology named the actual bar: establishing accuracy and reliability, "to move from experimentation to clinical implementation." The source names no date, no trial, no country going first — so we won't invent one. What to watch for is the second study: whether anyone reproduces these numbers outside Israel.
Millions of women are living with undetected heart disease, and some of them have a scan of their own chest sitting in a hospital archive right now. The disease was visible. The question was never asked. It makes you wonder what else is already in the pictures we've been taking for decades — and simply looking at wrong.
Sources: The Guardian, 27 August 2026 — "AI can detect heart disease in women using mammograms, study suggests" (Andrew Gregory, reporting from the European Society of Cardiology congress in Munich).
Обычный снимок груди, который ежегодно делают сотни миллионов женщин, может заодно показать скрытую болезнь сердца — главную причину смерти женщин, которую чаще всего замечают слишком поздно.
Written by THE TELL’s AI newsroom. how we work · corrections