Diagnostic prediction of ulcerative colitis using simplified endoscopic features and blood count-derived indices: a multicenter study

Introduction Fecal calprotectin (FC) results may be unavailable during assessment of inflammatory or ulcerative intestinal lesions. We assessed FC documentation and developed a diagnostic prediction model using endoscopic features and blood-derived inflammatory indices for ulcerative colitis (UC). Methods We retrospectively included 655 patients from four centers. Four prespecified logistic models were compared in complete cases ( N = 455; 103 UC cases). Performance included discrimination, Brier score, calibration, and decision curves. Internal validation used 1,000 bootstrap resamples; transportability was assessed by leave-one-center-out cross-validation. LASSO-based analyses were exploratory; multiple imputation also assessed the robustness of the four primary models. Results FC was documented in 6 patients (0.9%). A model combining mucosal granularity, mucosal edema, neutrophil-to-lymphocyte ratio, and systemic immune-inflammation index had an apparent area under the receiver operating characteristic curve (AUC) of 0.739 (95% CI 0.684–0.795) and an optimism-corrected AUC of 0.733. It exceeded the blood-derived index model (AUC=0.594, P < 0.001) and simplified endoscopic model (AUC = 0.711, P = 0.024), but did not differ significantly from the three-item endoscopic model comprising vascular pattern abnormality, bleeding, and erosions/ulcers (AUC = 0.758, P = 0.521). At a cutoff of 0.155, Negative predictive value was 0.901. LASSO selected vascular pattern abnormality, bleeding, and mucosal granularity (optimism-corrected AUC = 0.746). Discussion The combined model showed moderate discrimination, with modest incremental value from blood-derived indices.