Overview
- Mayo Clinic researchers validated an AI model on nearly 2,000 CT scans that identified 73% of pancreatic cancers a median 16 months before diagnosis.
- The system, called REDMOD, reads subtle tissue patterns that are invisible to the human eye and outperformed expert radiologists who caught 39% of the same prediagnostic cases.
- Multi-hospital testing showed the tool works across different scanners, and a study named AI-PACED now evaluates how real-time alerts guide follow-up and limit false positives.
- Other teams report progress on screening tools, including a USC algorithm that finds a few cancer cells in blood in about 10 minutes and a Washington University model that refines five-year breast cancer risk from mammograms.
- Editors and experts say broader use depends on computing capacity, strong clinical validation, and financing, with cloud options and clear workflows helping clinicians use AI safely.