short
- Mayo Clinic’s REDMOD AI model achieved a 73% detection rate for pancreatic cancer in routine CT scans an average of 16 months before clinical diagnosis.
- The AI analyzed nearly 2,000 CT scans that were originally interpreted as normal, including scans of patients who were later diagnosed with pancreatic cancer.
- REDMOD’s detection capacity is nearly double that of specialists reviewing the same scans – 39% for radiologists versus 73% for AI.
Mayo Clinic has developed an artificial intelligence model that can detect pancreatic cancer up to three years before clinical diagnosis by identifying subtle changes on routine CT scans, according to the Mayo Clinic website. Validation study results published in the magazine, gut.
The AI model, called REDMOD, detected pancreatic cancer an average of 475 days before clinical diagnosis with a specificity of 88%. The system correctly identified patients who did not develop cancer while detecting what researchers described as an “invisible” marker of preclinical pancreatic ductal adenocarcinoma.
“The biggest barrier to saving lives from pancreatic cancer is our inability to see the disease when it is still curable,” said Ajit Goenka, a radiologist and nuclear medicine specialist at Mayo Clinic, and lead author of the study. “This AI can now recognize a cancer signature from a pancreas that appears normal, and it can do so reliably over time and across diverse clinical settings.”
For cases detected more than two years before clinical diagnosis, the REDMOD system demonstrated three times the accuracy of radiologists, achieving 68% accuracy compared to 23% for human specialists reviewing the same imaging.
Mayo’s achievement comes amid broader advances in artificial intelligence in pancreatic cancer detection. PanDx, an AI framework for analyzing contrast-enhanced CT scans, recently achieved first place in the PANORAMA Challenge with an AUROC of 0.9263.
Mayo Clinic researchers are now conducting AI-PACED, a prospective clinical study evaluating how doctors can incorporate AI-guided detection into the care of high-risk patients. The study aims to translate the laboratory success of REDMOD into real-world clinical applications.
Early detection remains critical for pancreatic cancer, which is expected to become the second leading cause of cancer-related death in the United States by 2030. The same report noted that the disease’s lethality stems from its delayed onset, with more than 85% of patients receiving a diagnosis after the cancer has already spread.
It is expected that about 67,530 Americans will be diagnosed with pancreatic cancer in 2026. Current treatment options remain limited once the disease progresses beyond the pancreas. REDMOD grew out of Mayo Clinic’s Precure Initiative, a program focused on predicting and preventing disease by identifying early biological changes before symptoms appear.
Daily debriefing Newsletter
Start each day with the latest news, plus original features, podcasts, videos and more.
