Saw a video about a system being "able to detect" certain cancers "1.5 years earlier than human experts". At the very end: "It's not ready for real-world applications yet, as the false-positive rate is too high."
Sorry, but that's not "detecting" anything "earlier" (yet) then, it just has a low(er) threshold. ๐ E.g. I can easily "detect" skin cancer with higher accuracy than experts by simply categorising everything as skin cancer.
#Statistics really needs to become basic knowledge.