Cure8 research brief
Automated, consistent scoring of the UCEIS could reduce subjectivity and variability in endoscopic assessment of ulcerative colitis, which matters for clinical trials, central reading, and potentially treatment decisions if linked to outcomes.
Gastroenterologists and endoscopists, IBD researchers and clinical trial teams, and patients interested in how AI may improve objective measurement of disease activity.
This Scientific Reports paper reports UC-MTLNet, a multi-task deep learning model trained on >10,000 white-light endoscopic images from multiple centers to predict UCEIS descriptors, total UCEIS score, endoscopic remission, and severity strata.
The authors evaluated performance on internal and external test cohorts and compared the model to experienced endoscopists, finding higher exact-score accuracy and weighted-kappa for image- and patient-level UCEIS prediction under standardized image-based conditions.
The model achieved high agreement metrics (quadratic-weighted kappa values and accuracy for remission and severity strata) in the cohorts reported, and the authors highlight efficiency gains versus single-task models (fewer parameters and faster inference).
The paper frames UC-MTLNet as a potential tool to assist standardized UCEIS assessment in research settings and central reading, but notes prospective video-based and outcome-linked validation are still needed.
This is a diagnostic/methods study using still endoscopic images with internal and external test cohorts. The authors state that prospective video-based validation and linkage to clinical outcomes are needed before clinical adoption. The article is an open-access peer-reviewed journal report.
Review the original publication for the complete reporting, methods, and context.
Funding disclosed by the source: Shandong Provincial Medical and Health Science and Technology Program, award 202503031186
This Cure8 brief is based on source text from the linked article. Cure8 is informational only and is not a substitute for professional medical advice, diagnosis, or treatment.
Measuring both how severe inflammation is and how much of the colon is affected could give a fuller picture of disease burden than current focal endoscopic scores. For patients with UC, that may affect risk assessment...
A more detailed, reproducible measure of endoscopic inflammation could better reflect total disease burden and reduce variability from human readers. That may matter for monitoring disease activity, judging treatment...
Kidney dysfunction can occur in people with IBD and may be under-recognized. A validated prediction model could help clinicians identify IBD patients at higher risk and prompt earlier monitoring or interventions.
