Researchers at the Mayo Clinic have created an artificial intelligence system that can identify pancreatic cancer long before it is usually found, raising the possibility of earlier treatment for one of the most lethal forms of cancer.

The new technology, known as the Radiomics-based Early Detection Model (REDMOD), is designed to analyze standard abdominal CT scans and detect early warning signs of pancreatic cancer as much as three years before a formal diagnosis. According to researchers, the system evaluates hundreds of subtle imaging features related to tissue texture and structure, allowing it to recognize faint biological changes that are not yet visible as tumors.

In findings published in the journal Gut, the AI model successfully flagged 73% of cancers before diagnosis, with a median lead time of roughly 16 months. That performance is nearly twice as effective as specialists reviewing the same scans without the assistance of AI.

When looking at scans taken more than two years prior to diagnosis, the system proved even more effective, identifying almost three times as many early-stage cancers as human reviewers.

To ensure reliability, researchers tested the model on close to 2,000 CT scans collected from different hospitals, imaging systems, and clinical settings. The AI’s results remained consistent even when analyzing multiple scans from the same patient over time, suggesting it could be useful for continuous monitoring and early detection.

Another advantage of the system is that it operates automatically, eliminating the need for time-consuming manual input by clinicians.

This work is part of a larger push at Mayo Clinic to improve early detection of pancreatic cancer, a disease expected to become the second-leading cause of cancer-related deaths in the United States by 2030.

The AI tool is intended to evaluate scans that patients are already receiving for unrelated reasons, particularly in individuals considered higher risk, such as those who develop new-onset diabetes. It can flag elevated cancer risk even before any visible mass appears.

Researchers are now moving forward with clinical testing through a study known as AI-PACED, which will examine how doctors can incorporate AI-assisted detection into care for high-risk patients. The trial will track early detection rates, false positives, and patient outcomes.

According to a press release, the research is part of Mayo Clinic’s Precure initiative, which “aims to predict and prevent disease by identifying the earliest biological changes in the body before symptoms start.”

Pancreatic cancer remains especially dangerous because it typically produces no symptoms in its early stages. More than 85% of patients are diagnosed only after the disease has already spread, and the five-year survival rate remains under 15%, according to the National Cancer Institute.

“This AI can now identify the signature of cancer from a normal-appearing pancreas, and it can do so reliably over time and across diverse clinical settings,” said Ajit Goenka, the study’s senior author and a radiologist and nuclear medicine specialist at Mayo Clinic.

Detecting pancreatic cancer earlier could allow more patients to receive potentially curative treatment, improving outcomes for a disease that is often diagnosed too late.

Researchers say the model’s ability to function across different hospitals and imaging systems could allow it to be widely adopted, potentially benefiting patients nationwide.