Medical AI detects deadly pancreatic cancers up to 3 years sooner

A new artificial intelligence model developed by the Mayo Clinic in Rochester, Minnesota, can help specialists detect pancreatic cancer up to three years before most clinical diagnoses, allowing curative treatment.

The AI identifies subtle signs of disease before tumors are visible on routine abdominal CT scans, Mayo reports.

"The greatest barrier to saving lives from pancreatic cancer has been our inability to see the disease when it is still curable," senior author and Mayo Clinic radiologist Ajit Goenka said in a Mayo release. "This AI can now identify the signature of cancer from a normal-appearing pancreas, and it can do so reliably over time and across diverse clinical settings."

Researchers fed the AI model nearly 2,000 CT scans, including those from patients later diagnosed with pancreatic cancer, Mayo's website states. All the scans were originally categorized as normal, showing no cancer. The Radiomix-based Early Detection Model identified 73% of those patients with cancer at an average of 16 months before they were eventually diagnosed.

On average, the model could have doubled the number of cancers caught earlier. The earlier a patient could get screened, however, the greater the advantage they found. In scans obtained two years or more before diagnosis, the AI identified nearly three times as many early cancers that would otherwise go undetected.

The findings, published in Gut, follow a years-long effort by Mayo researchers to enable earlier detection of one of the deadliest cancers.

The failure to find pancreatic cancer early makes it incredibly deadly, according to the National Institutes of Health's National Cancer Institute. More than 85% of patients are diagnosed after the disease has already spread.

Although it makes up 3% of all cancers, with an estimated 67,000 new cases per year, the five-year survival rate after diagnosis is under 14%. Less than 2% of Americans will develop pancreatic cancer, but by 2030, NIH estimates it will become the second leading cause of cancer-related death.

Researchers reported that the model is designed to run automatically on CT scans obtained for any reason, and flag elevated risk before human doctors can see a growth. The results are remarkably stable, they wrote, providing consistent results months apart, supporting its use for routine detection and tracking progress over time.

_____

Copyright 2026 Tribune Content Agency. All Rights Reserved.