Journal of pediatric gastroenterology and nutrition
Cure8 research brief
Why This Matters
Predicting early need for infliximab could help identify children with Crohn’s disease who may benefit from earlier biologic therapy instead of prolonged conventional treatment. A reliable prediction tool may shorten time to effective treatment and reduce complications from delayed escalation.
Who Should Pay Attention
Pediatric gastroenterologists, clinical researchers in IBD and medical AI, and families of children newly diagnosed with Crohn’s disease interested in treatment planning and risk stratification.
Study Snapshot
Story typeResearch paper
Evidence typeResearch paper
Source depthJournal abstract
What To Know
A multicenter study developed and externally validated a machine-learning model that combines colonoscopy images and clinical data from children with newly diagnosed Crohn’s disease to predict whether a patient will start infliximab within 6 months after beginning conventional therapy.
The model used an attention-based multiple-instance learning approach with a ResNet-50 imaging encoder (pretrained on GastroNet-5M) plus a multilayer perceptron for clinical variables. Performance in the discovery cohort was AUROC 0.839; external-validation AUROCs were 0.857 and 0.688.
Explainability methods (attention maps and SHAP) highlighted inflamed mucosa/ulcers on images and clinical predictors including IgG, BMI, prealbumin, hematocrit, and age.
This work suggests multimodal AI could aid early risk stratification to identify pediatric patients likely to need early biologic (infliximab) therapy, but it is an initial study with modest external sample sizes and needs broader validation before clinical use.
Keep In Mind
Structured content depth: abstract. The study reports model development and external validation but with small external cohorts (n=10 and n=20), so results are preliminary and require larger, diverse validation before clinical deployment. This summary is based on the article abstract.
Source Details
Review the original publication for the complete reporting, methods, and context.
Research paper
Evidence type derived from source or registry metadata.
PublicationJournal of pediatric gastroenterology and nutrition
AuthorsSeon Young Kim, Seung Ha Hwang, Dongjin Yeo +7 more
InstitutionDepartment of Pediatrics, Kyung Hee University Medical Center, Kyung Hee University School of Medicine, Seoul, South Korea.
Study typeJournal article
Indexed viaPubMed
Source typeResearch paper
PublishedOct 15, 2026, 12:00 AM
Content availableJournal abstract
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.
Pharmacometric Targets During Anti-TNF Induction and Their Association with Deep Remission in Pediatric Crohn’s Disease
The study identifies specific induction exposure targets and biomarkers tied to higher chances of deep remission at 12 months in children on anti-TNF therapy, which could inform earlier dose optimization to improve...
Autologous Stem Cell Transplant for Crohn's Disease
This trial is testing whether adding vedolizumab after autologous stem cell transplant can improve remission rates for people with severe Crohn's disease that has failed multiple therapies.
Gradient Boosting Machine-Based Prognostic Model with SurvSHAP Analysis for Predicting Secondary Loss of Response to Infliximab Therapy in Crohn's Disease.
The work identifies factors that predict loss of response to infliximab and suggests maintenance drug levels are the strongest predictor, which could influence monitoring strategies for people with Crohn's disease on...