Table 3. Apparent and cross-validation performance of the logistic regression models.

	Model 1	Model 2	Model 3
Patient group	Group A	Group A	Group A
Control group	Group B + Group C	Group B	Group C
Penalty (alpha)	0.3	0.3	0.3
Apparent performance
AUC	0.82	0.75	0.96
Cut-off	0.20	0.30	0.30
Sensitivity, %	98.6	94.3	100
Specificity, %	33.0	20.3	78.4
NPV, %a	99.3	95.2	100
Cross-validation performance
AUC	0.78	0.65	0.94
Cut-off	0.20	0.30	0.30
Sensitivity, %	95.4	91.3	97.3
Specificity, %	35.0	27.6	69.1
NPV, %a	97.7	94.7	99.3

AUC, area under the receiver operator characteristic curve; NPV, negative predictive value

Group A, patients; Group B, controls; Group C, young healthy volunteers.

^aBased on a disease prevalence of 15%.