A new AI tool called Mia has shown it can spot signs of breast cancer that doctors miss.
In a test run with UK NHS doctors, Mia checked over 10,000 women’s mammograms. Most of the women didn’t have cancer, but Mia found all the cases that did, plus 11 more that the doctors didn’t catch.
Mia was trained on data from over 6,000 past breast cancer cases to learn what cancer looks like in scans. When put to the test, Mia got it right 81.6% of the time when cancer was present, and 72.9% of the time when it wasn’t.
Breast cancer is a big deal for women around the world, with millions of new cases each year. Even though treatment is getting better, it can still have tough side effects.
Now, researchers are working on teaching Mia to predict which patients might have these side effects after treatment. This could help doctors give better, more personalized care to those who need it most.
They’re planning a big trial with 780 breast cancer patients to see how well Mia’s predictions hold up over a couple of years. The hope is that eventually, Mia will be able to give doctors a full picture of a patient’s health and what kind of treatment they might need.
- UK and US Forge Agreement to Create AI Safety Tests - April 2, 2024
- UAE Setting Up AI-focused fund to Build its Own Chip - March 25, 2024
- New AI Tool Detects Cancer Signs Radiologists Missed - March 25, 2024