Table 5. AUC (mean) for clinical scores and trained radiomics models.

	PSA density	Mean ADC	PI-RADS	Regression	SMOTE mRMR(25)	cw | mRMR(25)	cw | no FS	no cw | mRMR(25)	cw | mRMR(50)	cw | mRMR(100)
AUC (mean)	0.63	0.71	0.78	0.80	0.78	0.80	0.82	0.80	0.81	0.83
AUC [1]	0.73	0.77	0.85	0.91	0.88	0.92	0.88	0.91	0.91	0.92
AUC [2]	0.63	0.77	0.74	0.76	0.64	0.68	0.67	0.71	0.70	0.71
AUC [3]	0.63	0.65	0.87	0.71	0.84	0.84	0.89	0.83	0.85	0.87
AUC [4]	0.53	0.59	0.68	0.77	0.64	0.65	0.83	0.69	0.74	0.82
AUC [5]	0.64	0.78	0.75	0.86	0.88	0.89	0.84	0.87	0.84	0.81
P values	<0.001	0.003	0.185	0.478	0.186	0.436	0.953	0.496	0.610	1

For each score or configuration, the mean AUC and the result for each CV fold is shown. The bottom row shows the P values yielded by the mixed-model analysis, showing how significantly different the classification performance of each of the model configurations in comparison to the best-performing “cw | mRMR(100)” was. Model configuration shorthands according to Table 1. AUC, area under the curve; PSA, prostate specific antigen; ADC, apparent diffusion coefficient; PI-RADS, Prostate Imaging Reporting and Data System; SMOTE, synthetic minority oversampling technique; mRMR, minimal redundancy maximal relevance; cw, class weights.