Study: The US develops AI model that can detect pancreatic cancer 16 months before official diagnosis

Researchers at the Mayo Clinic and the University of Texas Cancer Center have developed an artificial intelligence model, REDMOD, that can detect pancreatic cancer 16 months before the official diagnosis.

This is stated in a study in the journal Gut, writes ScienceAlert.

The model was initially trained on 969 computed tomography (CT) scans of the pancreas to learn to detect subtle signs of cancer in its early stages.

After training, the AI analyzed the CT scans of 430 healthy people and 63 patients who were later diagnosed with cancer. Of the 63 patients, it suspected cancer in 46 people (73%) about 16 months before they were officially diagnosed. Some were diagnosed more than two years before they were officially diagnosed.

Also, one scan of the model was given for analysis several times — most often it gave the same result. The AI model does not look for usual tumors, but for violations of the texture and structure of the tissue. Next, they want to test it on wider groups of people.

• Another major cancer study was released in October 2025. The Galleri test, manufactured by the American pharmaceutical company Grail, was able to detect fragments of cancer DNA that had broken away from tumors and were circulating in the blood. The blood test for more than 50 types of cancer could help speed up diagnosis.

For more news and in-depth stories from Ukraine, please follow us on X.