In a promising leap for cancer diagnostics, scientists in the United States have developed an artificial intelligence system capable of detecting pancreatic cancer years before traditional methods. This
innovation could be a game-changer for a disease that remains one of the toughest and deadliest cancers to catch in its early stages.
The AI tool, dubbed REDMOD, is the result of collaborative research by teams at the Mayo Clinic and the University of Texas MD Anderson Cancer Center. Unlike the human eye, which can miss subtle warning signs in medical imaging, REDMOD excels at analysing CT scans for minute changes in tissue texture and structure, clues that often go unnoticed by even highly skilled specialists.
The results are remarkable. According to the study, REDMOD identified early signs of the most common type of pancreatic cancer in about 75% of cases, on average 16 months before a formal diagnosis was made. In some patients, the AI system flagged abnormalities more than two years ahead of a human diagnosis. This is a significant improvement over conventional radiology, which detected the same signs at roughly half the rate.
To validate the tool’s accuracy, the researchers trained REDMOD on nearly 1,000 CT scans and then tested it on a separate group, including 63 scans from individuals who later developed pancreatic cancer. The AI correctly flagged 46 of these as suspicious, even though human radiologists had previously cleared them.
However, the system isn’t perfect yet. Among 430 healthy individuals, 81 were incorrectly flagged, which would mean extra testing and follow-ups for those patients. Despite this limitation, experts like Mayo Clinic radiologist Dr Ajit Goenka are optimistic, noting that REDMOD can " identify the signature of cancer from a normal-appearing pancreas and could dramatically improve detection rates.< p fetchpriority="low" loading="lazy">
Pancreatic cancer is projected to become the second leading cause of cancer-related deaths in the U.S. by 2030, largely because most cases are discovered only after the disease has spread. Earlier detection, as made possible by tools like REDMOD, could lead to curative treatments for many more patients.
Researchers caution that further studies are needed with larger and more diverse populations before the AI can be widely adopted in clinics.
