Plain English with Derek Thompson One of the Deadliest Cancers in America May Have Met Its Match
102 snips
May 5, 2026 
Dr. Ajit Goenka, Mayo Clinic radiologist studying AI for earlier cancer detection, explores a possible turning point for pancreatic cancer. He gets into AI spotting warning signs on CT scans years early. They also cover KRAS-targeting drugs, personalized mRNA vaccines, who should actually be screened, and why medical breakthroughs need slow, careful validation.
AI Snips
Chapters
Books
Transcript
Episode notes
Early Detection Could Unlock Every Other Advance
- Early detection may matter most because pancreatic cancer is usually found too late for treatment to help.
- Derek Thompson frames Mayo's AI as the third puzzle piece after drugs and vaccines because it may detect cancer on ordinary CT scans years earlier.
Mayo Tried To Keep The AI From Cheating
- Mayo trained AI on CTs taken three months to three years before diagnosis, aiming to detect invisible mathematical signals.
- Ajit Goenka says radiologists manually confirmed no visible cancer on each scan and matched controls carefully so the model could not simply memorize obvious clues.
Better Detection Still Fails If You Screen Everyone
- The model outperformed radiologists on earlier scans, but performance only matters alongside false-positive risk and disease prevalence.
- Ajit Goenka says even a near-perfect test creates many false alarms in low-risk people, so screening should focus on higher-risk adults over 50 with specific factors.
