[d0liver]

Yes, but then the study result should be, "AI correctly identifies 100% of breast cancers in study"

If we're saying there was a discrepancy and we're saying that all of the patients had cancer, then it would seem that there must have been some that were identified as not having cancer by AI.