Ulcerative colitis (UC) is a chronic, relapsing, immune-mediated disease characterized by colonic mucosal inflammation that leads to bloody stools, frequent bowel movements and abdominal pain.
Short-term treatment goals for UC involve alleviating disease symptoms, however long-term goals are targeted toward clinical remission.
Mucosal healing is a key predictor of long-term remission in UC. However, variability in histologic scoring among pathologists can limit development of targeted treatments.
Artificial intelligence (AI) offers a potential solution to reduce this variability.
Artificial intelligence-assisted histological scoring shows promise as a robust method for quantifying key histologic features
Good correlation between cellular Human Interpretable Features (HIFs) and RNAseq-derived cellular proportions in colonic tissue, validating two orthogonal techniques for determining relative cellular abundance in tissue
Strong concordance between pathologist-derived Geboes Scores and those derived from AIM-HI UC, particularly in Geboes Subgrades 2b and 3 that determine histologic improvement and remission status
Severely inflamed samples showed up to 46% discordance, potentially due to artifact overcall in area of erosion or ulceration with version 1 of the algorithm
Future work will evaluate potential improvements resulting from an upcoming version update for both algorithms.
Conference
Digestive Disease Week 2025
Partner
Gilead
View Poster