Smartphone apps that use artificial intelligence to assess skin cancer risk based on images of suspicious moles aren't ready for prime time, a recent systematic review in the BMJ suggests. The 9 studies included in the review "showed variable and unreliable test accuracy" for 6 such apps, 2 of which are approved by European regulators as medical devices. Of those 2 apps, only 1 was supported by published peer-reviewed studies, and its accuracy in those studies was poor compared with experts. The reviewers concluded that, overall, the 9 diagnostic accuracy studies were small and of poor methodological quality. Among other problems, clinicians rather than consumers usually selected which moles were assessed and took the pictures.