AI Should Replace Waiting, Not Doctors: Breast Screening in the Gulf

Every October, as a hospital chief executive, I supported breast cancer awareness and screening campaigns: pink ribbons in the lobby, free mammograms, talks in schools and workplaces. They mattered, and for a few weeks the clinics were busy. But the question I kept coming back to afterwards was a simple one. Who did we not see? The women who could not take a morning off work. The women with no one to drive them. The women told they were too young to worry. And the women who stayed away because they were afraid of the test itself.

A different patient, a borrowed model

Much of the screening model used in the region was designed around women in Europe and North America. The women it is meant to protect here look different.

In the Gulf, breast cancer is diagnosed about a decade earlier. Regional reviews put the median age at diagnosis at around 48 to 52, compared with around 60 in Western countries. Gulf cancer registry data show that roughly a quarter of cases occur in women under 40, the age at which most screening programmes have not yet started. Between a half and a little over half of women are diagnosed at an advanced stage.

The tissue is different too. A 2024 multicentre study in Riyadh found that about 43% of women aged 40 to 79 had dense breasts. Dense tissue makes a mammogram harder to read, and a recent Saudi study found ultrasound outperformed mammography for early detection in women under 50.

Then there is uptake. The last nationally representative survey across Kuwait, Oman, Saudi Arabia and the UAE found that only 5% to 15% of women aged 40 to 75 had had a mammogram. That survey is now more than fifteen years old. Uptake has certainly improved since, but no comparable regional figure has replaced it, and that gap in our knowledge tells its own story.

Put together, the problem is not that we lack a better detector. It is that the right scan does not reach the right woman at the right age.

Where AI actually helps

This is where I think AI earns its place: not by replacing the radiologist, but in three less glamorous jobs.

Capturing a good scan. Ultrasound works well in dense tissue. Published evidence suggests it finds roughly four to seven additional cancers per 1,000 women with dense breasts. But handheld ultrasound depends on a skilled sonographer, and they are scarce. A new category of automated 3D breast ultrasound uses AI to guide and check the scan as it happens. A trained nurse or medical assistant in a primary care clinic can then capture a complete examination, which a radiologist reads remotely. A good diagnosis starts with a good image.

Managing the volume. A full 3D breast scan produces thousands of images. No radiologist reads them all with the same attention at the end of a long day as at the start. AI can flag the frames that deserve a closer look and move urgent cases to the top of the list. The radiologist still makes the call.

Remembering. When a scan is taken the same way every time, this year's images line up with last year's. AI can then compare a woman with her own previous scans rather than with an average of millions of other women. For me, this is the heart of Connected Care: continuity between encounters, so that each visit builds on the last instead of starting again.

A word of caution. Many of these capabilities are still in development or under regulatory review. Hospitals should adopt them on evidence, not on enthusiasm.

Replacing waiting, not people

The Gulf does not have too many radiologists or sonographers. It has too few, and often not where women live. The real enemy is the wait: referral, appointment, travel, scan, report, result. Each step is a point where a woman can drop out, and each week is a week of worry.

That is where AI should do its work. Let software handle the repetitive, measurable tasks: guiding the scan, checking image quality, sorting the images and comparing them over time. Then the specialist's time goes where it matters most. The decision about a biopsy, and the conversation with a frightened woman and her family, must always stay with a doctor. AI can inform those moments. It should never own them.

Questions for boards and leadership teams

1. Where will the scan happen, and is it where women already go for care: primary care, women's health clinics, workplaces, mobile units?
2. Who reads the image, how quickly does the woman get an answer, and who gives it to her?
3. Has the AI been validated on women like ours, younger and with denser tissue, and approved by our regulator?
4. Will this year's scan be comparable with next year's?

The best screening technology is the one that reaches a woman before her cancer reaches a late stage.

Sources include a review of breast cancer in the Arabian Gulf countries, Al-Shamsi et al. in Frontiers in Oncology (2024), AlSaleh et al. in Gland Surgery (2024), the Turkish Journal of Surgery (2026), So et al. in the European Journal of Cancer Prevention (2019) and Rebolj et al. in the British Journal of Cancer (2018).

Related reading: Before you press “Enable AI” · People first, technology next