People with ulcerative colitis (UC) face a higher risk of developing colorectal cancer compared to the general population. Low-grade dysplasia (LGD), which refers to abnormal or precancerous lesions, can serve as an early indicator, but not all UC-LGD cases progress to cancer. This uncertainty makes it difficult for both clinicians and patients to decide on the best course of action, such as whether to continue regular monitoring or consider preventative surgery.
Researchers at the University of California San Diego have developed an artificial intelligence (AI) tool that works alongside biostatistical risk models to predict which UC-LGD patients are most likely to develop colorectal cancer. Their study, published in Clinical Gastroenterology and Hepatology on February 17, used medical records from 55,000 patients in the U.S. Department of Veterans Affairs health care system. This dataset is reportedly the largest of its kind in the United States.
The AI workflow reviewed colonoscopy and pathology reports to identify UC-LGD patients and assess their individual risk for developing cancer. According to Curtius, a research health scientist at VA San Diego Healthcare System, "A lot of people are low risk - they have small dysplastic lesions - and it's been hard to know what to confidently tell these people until now." Curtius explained that current recommendations suggest patients with small lesions return for surveillance every two years. "With this tool, there may be a potential to increase the surveillance interval so patients who are at this low risk don't have to come back so often."
The model also found that individuals with unresectable visible lesions—those that cannot be safely removed due to size, location or extent—are at much higher risk than many clinicians previously estimated.
The researchers believe this AI approach could integrate into existing clinical workflows by providing automated risk assessments for each patient. These assessments could help guide decisions about when patients should return for follow-up colonoscopies or consider surgical options while easing workload pressures on healthcare teams.
Curtius noted: "Currently, the process of advising people about levels of risk is a somewhat subjective thing, and doctors don't have enough data to back up what they feel. This AI pipeline could read the clinical notes and tell you your risk score, rather than just having a list of risk factors and no real way to turn that into a number during a patient visit."
The technology might also assist in identifying patients who need prompt follow-up appointments—a factor known to contribute significantly to preventing colorectal cancers when addressed without delay.
Future work will focus on validating this AI tool outside the VA system and integrating additional information such as new risk factors and genetic data from patients. As Curtius said: "We know that genomics play a big part in driving cancer progression."
