Predictive model for ulcerative colitis therapeutic response using clinical indicators and peripheral blood T-cell subsets.

OBJECTIVE: This study aimed to integrate clinical indicators with peripheral blood T-cell subsets to develop a prediction model for ulcerative colitis (UC), and to facilitate precise subtyping and individualized treatment decisions. METHODS: A retrospective cohort of 346 UC patients (June 2023-June 2025) was randomly assigned to training (n=242) and validation (n=104) sets at a 7:3 ratio. Univariate analysis, Least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression to identify independent influencing factors. Machine learning models-random forest (RF), gradient boosting, and logistic regression-were constructed using the selected core variables. Model discrimination, calibration, and clinical utility were assessed using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). RESULTS: According to week-14 response criteria, the training set comprised 158 responders (65.3%) and 84 non-responders (34.7%). Six indicators significantly associated with treatment response: albumin, C-reactive protein (CRP), tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), CD3+CD4+CD25+/CD3+CD4+ T-lymphocyte ratio, and CD4/CD8 ratio were as independent influencing factors. Among the models, RF exhibited the highest numerical AUC, with an AUC of 0.852 in the training set and 0.804 in the validation set. The calibration curve demonstrated good agreement between predicted and actual risks. DCA indicated higher net clinical benefit within a risk threshold range of 0.1-0.8. CONCLUSION: A predictive model for UC treatment response based on clinical indicators and peripheral blood T-cell subsets was constructed and validated, and may serves as an exploratory tool for individualized treatment decisions in UC.