How AI Is Spotting Pancreatic Cancer Signals That Most Doctors Miss

A model developed at the Mayo Clinic identified early signs of pancreatic cancer in routine scans more than a year before diagnosis.

BY LEILA SHERIDAN, NEWS WRITER

Artificial intelligence may be able to detect pancreatic cancer more than a year before it’s typically diagnosed, offering a potential breakthrough for one of the deadliest forms of cancer.

Researchers at the Mayo Clinic developed a model called Redmond that identified subtle warning signs in routine CT scans an average of about 475 days before patients were diagnosed, according to a study published Tuesday in the journal Gut.

The system works by analyzing patterns in imaging that are invisible to the human eye. It was trained and tested on more than 1,400 patients, including 219 people whose scans were initially read as normal but who later developed pancreatic cancer, Bloomberg reported.

In head-to-head comparisons, the AI significantly outperformed radiologists at spotting these early signals. It correctly identified 73 percent of cases, compared with about 39 percent for doctors reviewing the same images. The gap widened further for scans taken more than two years before diagnosis, where the model detected 68 percent of cases versus 23 percent for radiologists, according to Bloomberg.

Vibe-Coding for Beginners in Five Easy Steps

Pancreatic cancer is notoriously difficult to catch early. Tumors often don’t cause symptoms and typically don’t appear clearly on imaging until the disease is advanced. As a result, more than 85 percent of cases are diagnosed too late for curative treatment, helping explain why five-year survival rates hover around 10 percent globally, Bloomberg reported.

By identifying patients earlier, before tumors are visible or symptoms appear, the model could shift diagnosis from a late-stage reaction to a proactive process.

“This temporal window holds profound significance, as attaining such early detection would substantially augment the probability of cure and improved survival,” the researchers wrote.

© Inc.com