Cork scientists develop AI tool successful in diagnosing ulcerative colitis

1 min read

Scientists from the Cork research centre for food and medicine, APC Microbiome, have developed an artificial intelligence device (AI) that can successfully diagnose ulcerative colitis.

A recently joined professor from University College Cork (UCC) led a team of international researchers on a project that has developed and validated an AI system for the management of ulcerative colitis.

Professor Marietta Iacucci’s team created an AI computer-assisted diagnostic tool that speeds up the process of predicting outcomes in patients diagnosed with ulcerative colitis. The tool also minimises errors when it comes to evaluating biopsies in a clinical setting.

“This novel AI system can be of great help to pathologists and clinicians in identifying and distinguishing between inflammation and remission ­– the primary goal in ulcerative colitis treatment,” commented Iacucci.

“We believe [the system] is set to alter how ulcerative colitis clinical assessments and research trials are conducted,” she added. Ulcerative colitis is a chronic inflammatory bowel disease with no known cure and causes severe pain and ulcers in the digestive tract.

The team believes the tool can not only support the treatment of that condition, but also on other conditions where tissue is affected. Researchers tested the tool on 535 biopsies drawn from 273 patients spread across 11 international centres. Approximately two-thirds to three-quarters of the biopsies were in remission. The AI tool predicted endoscopic inflammation levels at 80pc accuracy when it was used to detect the presence of ulcerative colitis.

Its detection abilities could be better than traditional clinical methods, which are often difficult to train staff to get used to, expensive and inaccurate as different clinicians have different reads based on what they see in the tissue themselves.

The AI tool’s most notable feature is its ability to predict disease flare risk, an area in which it outperformed traditional diagnostic methods.