A review of 12 studies about AI systems used to screen patients for breast cancer found that they lacked evidence to prove that they would work well in a clinical setting. Researchers highlighted concerns about study quality, including the fact that all of the studies were retrospective and many of them had a high risk of bias.
