Table 3. Comparison of the performances of risk group predictions using benchmark methods and the proposed methods (scP.V and scP.W).

	PAM50	Mamma	RS	GGI97	Endo	LM	scP.V	scP.W
TCGA(OS)	1.45	0.95	1.33	0.87	0.87	1.43	1.76	1.32
TCGA(RF)	1.61	1.01	1.26	0.97	1.17	1.09	1.30	1.45
METABRIC(OS)	4.90	4.67	5.99	2.51	4.80	3.60	4.71	6.07
METABRIC(RF)	5.17	3.87	5.29	2.47	7.16	3.95	5.93	5.45
GEO	1.20	1.52	2.50	0.91	1.85	0.86	1.37	1.51
UK	1.54	1.72	1.62	1.68	2.17	1.74	2.90	3.27
Mean rank	4.83	3.33	5.33	1.58	5.25	3.33	6.00	6.33

The top-performing result is highlighted for each dataset. The reported hazard ratio is the average hazard ratio of 100 runs of 10-fold cross-validation on each dataset.