A Microsoft-powered medical AI spotted cancer in 11 women where doctors didn't
Getty Images
- A UK trial over a new medical AI found breast cancer in cases where human doctors missed it.
- The AI, which works on Microsoft's Azure cloud, found very small tumors in 11 women.
- Used more widely, Mia could improve cancer survival rates and slash the workload of doctors.
A medical AI tool helped identify signs of breast cancer in 11 women that doctors had overlooked, potentially heralding better, more efficient cancer screening.
The 11 cases were part of a mass trial of AI mammograms in the UK's National Health Service, which covered 10,889 patients, according to the BBC.
The AI, called Mia, was developed by the British firm Kheiron Medical Technologies.
The trial saw it employed by NHS Grampian, a health authority in Scotland, with data analyzed via Microsoft's Azure cloud network, per a Microsoft press release.
The press release said the tool allowed doctors to find 12% more cancers than what was typically expected.
It said that its use could cut the medical workload of the scans by 30%, making the process faster and more efficient.
As well as being faster than humans, the AI appeared to pick up on signs that even highly-trained professionals did not.
The most striking finding was the 11 women where the AI saw tumors too tiny for trained radiologists to notice.
One of them, mentioned in the Microsoft press release, was a woman called Barbara from the Scottish city of Aberdeen. (It didn't give her surname.)
"My cancer was so small that the doctors said it would not have been picked up by the human eye," Barbara said.
She described Mia as "a life saver." The BBC said her 6mm tumor only required minimal surgery and five days of radiotherapy.
Per the BBC, there were no cases where the human doctors noticed a cancer case that the AI had missed.
"If you pick up cancer under 15mm, most women now will have a 95% survival rate," said Dr. Gerald Lip, who led the study.
"Not only did Mia help us find more cancers, most of which were invasive and high grade, but we also modeled that it could reduce the time it takes to notify women from 14 days to just 3 days, reducing significant stress and anxiety for our patients," he said.
The tool could change the way medical services work with cancer patients in the long term.
Professor Lesley Anderson, Chair in Health Data Science at the University of Aberdeen, said the tool would significantly improve breast screenings as more cancers would be detected without having to ask patients to take part in additional tests.
The trial was funded by the UK government, and caught the attention of its most senior ministers.
Jeremy Hunt, Britain's finance minister, wrote on X that "innovations like this that can drive huge improvements in productivity in our public services."