Table 3.

Discrimination power of the prediction models

Models	AUC	Low 95% CI	High 95% CI	Optimal cutoff value	Sen	Spe	Acc	PLR	NLR	DOR	PPV	NPV
1 Stepwise	0.82	0.76	0.88	−0.1052	0.76	0.8	0.78	3.81	0.29	12.84	0.77	0.79
2 BS-stepwise	0.8	0.74	0.87	−0.1630	0.78	0.74	0.76	3.08	0.28	10.79	0.73	0.79
3 mfp	0.81	0.74	0.87	−0.0718	0.71	0.76	0.74	3.05	0.37	8.14	0.73	0.75
4 Full	0.83	0.77	0.89	0.0349	0.75	0.81	0.78	3.97	0.3	12.88	0.77	0.78
5 BS-full	0.83	0.77	0.89	−0.7773	0.83	0.71	0.77	2.89	0.22	12.68	0.72	0.83

AUC, area under the curve; AIC, Akaike Information Criterion; BS, bootstrapping; CI, confidence interval; DOR, diagnostic odds ratio; mfp, Multiple Fractional Polynomial; NPV, negative predictive value; PLR, positive likelihood ratio; PPV, positive predictive value; NLR, negative likelihood ratio; Sen, sensitivity; Spe, specificity.

¹Backward stepwise variables selection based on AIC.

²Stepwise model Bs 1,000 times.

³Final model was generated based on mfp.

⁴Full model.

⁵Full model Bs 1,000 times