Table 4. Bayesian multivariable logistic regression models of escalation to anti-TNFα therapy by week 52 for patients with moderate-to-severe disease.

	All patients in clinical model	Patients with biological data
	(n = 386; 147 [38%] events)	(n = 177; 69 [39%] events)
		Clinical model	Clinical and biological model
Baseline predictors
PUCAI score <45	1.80 (1.20, 2.90)	-	-
Haemoglobin ≥10 g/dL (without week 4 remission)	5 (1.8, 17)	7.3 (1.8, 42)	6.34 (1.44, 38.82)
Week 4 remission	11.20 (4.10, 37.00)	16.7 (4.3, 93)	17.25 (4.16, 101.17)
Antimicrobial peptide gene signature	-	-	0.55 (0.38, 0.80)
Ruminococcaceae (560535) OTU log relative abundance	-	-	1.45 (1.04, 2.07)
Sutterella (589923) OTU log relative abundance	-	-	0.80 (0.64, 0.99)
Model evaluation
AUC	0.69 (0.68, 0.70)	0.68 (0.68, 0.68)	0.75 (0.72, 0.76)
CV-AUC	0.68 (0.66, 0.71)	0.67 (0.58, 0.68)	0.73 (0.63, 0.81)
Sensitivity	0.38 (0.33, 0.67)	0.48 (0.00, 0.70)	0.51 (0.35, 0.65)
Specificity	0.82 (0.61, 0.86)	0.72 (0.60, 1.00)	0.78 (0.65, 0.91)
Positive predictive value	0.57 (0.52, 0.60)	0.37 (0.00, 0.53)	0.61 (0.53, 0.72)
Negative predictive value	0.69 (0.68, 0.75)	0.71 (0.61, 0.76)	0.72 (0.68, 0.75)
Clinical plus biological model vs clinical model§
Comparison of ELPD with SE			6.2 (4.1)

AUC = area under the curve. CV-AUC = 10-fold cross validation. §Comparison of the clinical plus biological model with clinical model in the subset of patients with biological data.