Researchers at the Mayo Clinic have developed an artificial intelligence model that detects pancreatic cancer on CT scans up to three years before clinical diagnosis, as reported in a study published in Gut. The AI identifies "subvisual" patterns in pancreatic tissue invisible to the human eye, analyzing radiomic features like minute textural disruptions. In comparisons with board-certified radiologists, it achieved 73% sensitivity on a test set of 493 scans, nearly double the 39% sensitivity of human doctors, and was almost three times more sensitive for scans taken over two years prior. This breakthrough could improve early detection for a disease with a 13% five-year survival rate, given the difficulty of spotting tumors due to the pancreas's deep location and lack of routine screening. However, the tool remains under evaluation in a clinical trial and is not yet ready for widespread use, though it may eventually triage high-risk individuals.