Mayo Clinic AI spots pancreatic cancer up to 3 years early

Mayo Clinic researchers report that REDMOD, an artificial intelligence model, identified pancreatic cancer on routine CT scans up to three years before clinical diagnosis, according to a validation study published in the journal Gut.

The team applied REDMOD to nearly 2,000 CT scans that had been originally read as normal, including images from patients later diagnosed with pancreatic ductal adenocarcinoma. The model detected cancer an average of 475 days before clinical diagnosis, with a median lead time of about 16 months, and produced a specificity of 88 percent.

REDMOD identified 73 percent of cancer cases in the sample, compared with 39 percent detected by specialist radiologists reviewing the same scans. For cancers detected more than two years before clinical diagnosis, the model achieved 68 percent accuracy versus 23 percent for radiologists. Researchers said the AI recognizes a subtle, “invisible” signature of pre-clinical disease on otherwise normal-looking pancreatic CT images.

REDMOD was developed under Mayo Clinic’s Precure initiative, a program that aims to detect early biological changes before symptoms appear. The validation study tested the model across different clinical settings to evaluate how reliably it could find early signs of pancreatic cancer on routine imaging not intended for cancer screening.

Mayo Clinic investigators are moving from retrospective analysis to prospective testing. The group has launched AI-PACED, a clinical study that will evaluate how REDMOD outputs might be integrated into care pathways for patients at elevated risk of pancreatic cancer and whether earlier AI-guided detection can translate into earlier intervention and improved outcomes.

The REDMOD results join other AI work on pancreatic cancer detection. A separate AI framework, PanDx, recorded an area under the receiver operating characteristic curve of 0.9263 in a recent challenge for contrast-enhanced CT analysis.

Pancreatic cancer is often found at an advanced stage: more than 85 percent of patients receive a diagnosis after the disease has spread beyond the pancreas. Public health projections estimate pancreatic cancer will be the second-leading cause of cancer-related death in the U.S. by 2030. About 67,530 Americans are expected to be diagnosed with the disease in 2026.

“The greatest barrier to saving lives from pancreatic cancer has been our inability to see the disease when it is still curable,” Ajit Goenka, M.D., the study’s senior author and a Mayo Clinic radiologist and nuclear medicine specialist, noted. He added that REDMOD can identify the cancer signature from a normal-appearing pancreas across diverse clinical settings.