Mayo Clinic AI Detects Pancreatic Cancer Up to 3 Years Before Diagnosis — REDMOD Study

Mayo Clinic's Radiomics-based Early Detection Model (REDMOD), published in the journal Gut, analyzed nearly 2,000 routine abdominal CT scans and identified 73% of pre-diagnostic pancreatic cancers at a median of ~16 months before clinical diagnosis — nearly double the 39% detection rate achieved by specialist radiologists reviewing the same scans manually. At lead times beyond two years, REDMOD was nearly three times more sensitive than human review. The model analyzes approximately 1,000 radiomic features — including texture, density, and spatial patterns — using wavelet-filtered image analysis and three machine learning algorithms, without requiring any special scan preparation beyond standard abdominal CT. Pancreatic cancer carries a 5-year survival rate under 13% primarily because over 80% of cases are caught at late, surgically inoperable stages; early detection is the single most impactful intervention. A prospective clinical trial (AI-PACED) has been initiated to validate REDMOD in real-world screening workflows. The study adds to a growing body of evidence that AI radiomic analysis of routine scans can dramatically shift cancer detection earlier, potentially transforming outcomes for one of the deadliest cancers globally.