Table 3.

Predictive accuracy (area-under-the-curve) of kallikreins, with correction by repeated 10-fold cross-validation

Predictor	Laboratory Model		Clinical Model
	Any cancer	High-grade cancer	Any cancer	High-grade cancer
Base model	0.680 (0.636, 0.727)	0.816 (0.741, 0.881)	0.724 (0.677, 0.771)	0.868 (0.795, 0.925)
Base model+Free PSA	0.762 (0.727, 0.807)	0.832 (0.780, 0.916)	0.779 (0.746, 0.828)	0.867 (0.830, 0.941)
Base model+Intact PSA	0.688 (0.651, 0.740)	0.798 (0.760, 0.892)	0.724 (0.687, 0.774)	0.853 (0.822, 0.930)
Base model+hK2	0.725 (0.686, 0.772)	0.826 (0.771, 0.902)	0.760 (0.725, 0.809)	0.864 (0.803, 0.933)
Full model	0.832 (0.804, 0.874)	0.870 (0.841, 0.937)	0.836 (0.810, 0.880)	0.903 (0.860, 0.960)

The base model for the laboratory model includes age and total prostate specific antigen (PSA), and for the clinical model includes age, total PSA, and digital rectal examination result. The full model includes the base model plus free PSA, intact PSA, and human kallikrein 2 (hK2). Cancers with biopsy Gleason score 7 and higher were considered high grade. 95% confidence intervals are given in parentheses, and were calculated using bootstrap methods with 1000 replications.