Skip to playerSkip to main content
  • 14 hours ago
A study in the journal Gut reports that REDMOD flagged signs of pancreatic cancer on CT scans an average of 475 days before clinical diagnosis. Researchers tested the AI tool on 219 patients whose earlier scans looked normal to radiologists; it detected 73 percent of cases, compared with 39 percent for radiologists. In an independent group of 539 patients, REDMOD correctly classified about 81 percent of cancer-free scans. The study population lacked ethnic diversity, and prospective testing is still needed. The video covers the numbers behind the tool.

Disclosure: This video contains stock footage and content created or enhanced using AI-assisted tools.

Category

🗞
News
Transcript
00:00An AI tool spotted signs of pancreatic cancer, an average of 475 days before diagnosis.
00:06The tool is called RedMod, and it reads texture patterns on CT scans.
00:10Researchers studied 219 patients whose earlier scans had looked normal to radiologists.
00:16All of them were later diagnosed with pancreatic cancer.
00:19The scans were taken from three months to about three years before diagnosis.
00:24RedMod caught 73% of those cases, compared with 39% for radiologists.
00:30On scans taken more than two years earlier, it caught 68% versus 23%.
00:35The tool also correctly cleared about 81% of cancer-free scans in a group of 539 patients.
00:42The findings were published in the journal GOOT.
00:45Researchers say the study group lacked ethnic diversity, which may affect results.
00:50They also say testing in real-world high-risk groups is still needed.
00:54More research is needed before the tool can be used in routine care.
00:57Disclosure. This video contains stock footage and content created or enhanced using AI-assisted tools.

Recommended