Answers are grounded in this paper and cite their sources.
Background Inflammatory Bowel Disease is a group of chronic, non-specific intestinal inflammatory disorders with incompletely defined etiologies, mainly including Ulcerative Colitis and Crohn's Disease. Its global incidence and prevalence are on the rise, posing a prominent public health challenge. Currently, clinical diagnosis of IBD relies on a comprehensive combination of multiple examinations, but existing methods have limitations such as invasiveness, high cost, or insufficient sensitivity. Therefore, identifying simple, non-invasive, and economical biomarkers to evaluate discrimination between established IBD and healthy controls is of great clinical significance. Methods A retrospective development and internal-validation cohort and a prospective temporal-validation cohort were used. A total of 297 retrospective participants (176 IBD patients and 121 healthy controls) and 50 prospective participants (28 IBD patients and 22 healthy controls) were enrolled. Clinical and laboratory data of all participants were collected, including inflammatory markers, nutritional status markers, and other related indicators. An exploratory L1-penalized logistic regression analysis of ten candidate indicators was performed in the training set. Four prediction models (Logistic regression, SVM, KNN, and RF) were constructed using NAPR and fecal calprotectin as the two biomarkers of interest. The models were evaluated and validated using ROC curves, single-marker comparisons, DCA, and interventional Shapley analysis. Results Baseline analysis showed that compared with the control group, the IBD group had significantly higher levels of inflammatory markers (NAPR, fecal calprotectin, CRP, ESR, etc.) and significantly lower levels of nutritional indicators (albumin, hemoglobin, etc.). At the cross-validated L1 penalty, ESR, HB, Na, IgG, and IgM, but neither NAPR nor fecal calprotectin, had nonzero coefficients. Among the four models, the RF model exhibited the best diagnostic performance, with an internal validation AUC of 0.905 and a prospective temporal-validation AUC of 0.881. In internal validation, the combined model had a higher AUC than either single-marker model; this advantage over fecal calprotectin alone was not evident in the prospective cohort. DCA suggested potential net benefit over a limited threshold range in the selected prospective cohort. In the two-marker RF model, mean absolute interventional Shapley values were higher for fecal calprotectin than for NAPR in internal validation. Conclusion The combination of NAPR and fecal calprotectin provided effective discrimination between established IBD and healthy controls in internal and prospective validation. This readily available biomarker combination offers a promising basis for further evaluation as a non-invasive adjunct to IBD assessment.
No takes yet. Share an insight, caveat, or question.
Lin et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: