AI reads mammograms for heart disease, not just breast cancer

Key points

• Researchers used AI to analyse routine mammograms, scans normally used only to check for breast cancer, and found signs of heart disease in the images.
• The AI identified women with coronary heart disease (a narrowing of the arteries supplying the heart) or high blood pressure, and those who had suffered a stroke.
• Heart disease is the leading cause of death in women globally and is frequently diagnosed late or missed entirely.
• No extra scan or appointment would be needed if the approach were rolled out, because it works on imaging women already receive.

Every year, millions of women sit in a radiology suite and have a mammogram. The scan takes a few minutes. Its job, as far as most women know, is to find breast cancer.

Researchers now say those same images contain something else: clues about the heart.

A study reported by The Guardian found that an AI model, software trained to spot patterns in medical images, could examine standard mammograms and correctly flag women who had coronary heart disease, high blood pressure, or had previously suffered a stroke. The researchers weren't modifying the scans or adding any extra step. They fed existing images into the AI and let it find signals that human eyes had never been asked to look for. Our 4 August story on PRISM2, a model that reads cancer images across multiple disease types, showed how quickly this class of multi-purpose diagnostic AI is moving.

Why does this matter for women specifically?

Heart disease kills more women than any other condition, yet it's still widely underdiagnosed in them. Women's symptoms can differ from the chest-clutching picture most people associate with a heart attack, so warning signs get missed.

Routine screening already reaches a large share of women in many countries. An AI layer running quietly alongside the standard cancer check could mean women at risk of a heart attack or stroke get referred to a cardiologist, a heart specialist, before a crisis happens. That's the promise here, and it's a meaningful one. But it's an early one.

What do we actually know, and what is still unproven?

The details needed to judge this fairly are thin. The Guardian's report doesn't specify the study's size, the institution behind it, or whether the results have appeared in a peer-reviewed journal, one where independent scientists vet the work before publication. Without those details, this sits closer to "promising research finding" than "proven clinical tool".

Accuracy figures haven't been published in full: how often the AI was right, and how often it flagged a problem that wasn't there. False positives, incorrect alerts in healthy women, could cause genuine anxiety and unnecessary follow-up tests. Trials in larger, more diverse populations will be essential before any health service could responsibly adopt this.

What happens next?

If further studies confirm the findings, health systems would need prospective trials, tests where the AI runs alongside real screening programmes and researchers track outcomes for the women it flags. Regulators would then need to approve it as a clinical tool. That path takes years.

For now, women should attend their scheduled mammograms and raise cardiovascular concerns directly with their doctor. This research doesn't change today's appointments. It points at what future ones might offer.

Common questions

Does this mean my next mammogram will check my heart?

Not yet. This is research, not a live clinical service. No health system has approved or rolled out this AI tool as standard care.

Should women worried about heart disease do anything differently right now?

Yes: speak to a GP or primary-care doctor about cardiovascular risk factors such as blood pressure and family history. Those conversations can happen today, without waiting for any new technology.