Mayo Clinic has developed an AI model capable of detecting pancreatic cancer up to three years before clinical diagnosis by analysing routine CT scans for subtle, previously invisible changes.
The REDMOD system identified cancer an average of 475 days before diagnosis with 88% specificity, analysing nearly 2,000 scans initially deemed normal.
“The greatest barrier to saving lives from pancreatic cancer has been our inability to see the disease when it is still curable,”
Said Ajit Goenka, radiologist at Mayo Clinic.
“This AI can now identify the signature of cancer from a normal-appearing pancreas.”
The model achieved a 73% detection rate compared with 39% for radiologists reviewing the same scans, and reached 68% accuracy versus 23% for specialists in cases identified more than two years early.
The research highlights AI’s growing role in medical diagnostics, particularly for diseases like pancreatic cancer that are often detected too late for effective treatment.
Mayo Clinic is now conducting further clinical trials to assess how the technology can be integrated into real-world healthcare settings for high-risk patients.
With pancreatic cancer projected to become the second-leading cause of cancer deaths in the US by 2030, earlier detection could significantly improve patient outcomes and survival rates.