A new AI tool developed by UC San Diego researchers uses clinical data to predict colorectal cancer risk in ulcerative colitis patients, helping clinicians make evidence‑based care decisions and preventing treatment delays.
Key Takeaways and AI Workflow Development
People with ulcerative colitis (UC) have an elevated risk of colorectal cancer, but determining which low-grade dysplasia (LGD) cases will progress to cancer has been a challenge for clinicians. Researchers at UC San Diego developed an AI workflow combined with biostatistical risk models to accurately predict cancer risk in UC-LGD patients. This automated system analyzed medical records from 55,000 patients in the U.S. Department of Veterans Affairs (VA) health care system, identifying individual cancer risks. The AI successfully grouped patients into five risk categories based on factors like dysplasia size, lesion resection completeness, number of dysplastic sites, and inflammation severity, demonstrating high accuracy over a decade. It also classified nearly half of the patients into the lowest-risk group, predicting a 99% chance of remaining cancer-free for two years. Additionally, the AI revealed that patients with unresectable visible lesions face a significantly higher risk than typically estimated.
A Boon to Patient Care
The AI model is designed to integrate seamlessly into clinical workflows, providing precise, automated risk assessments that can guide decisions for both clinicians and patients. This includes optimizing the timing of follow-up colonoscopies and considering when preventative surgery might be necessary. By offering objective risk scores derived from clinical notes, the tool moves beyond subjective assessments, reducing the burden on care teams and minimizing delays in essential follow-up care. The next phases of research will involve validating this AI tool in patient populations outside the VA system and incorporating emerging risk factors, including patient genetic information, to further enhance its predictive capabilities.
