Survival Prediction and Feature Selection in Patients with Breast Cancer Using Support Vector Regression

Table 3

Performance measures of SVR and Cox models for simulated experiments with different number of features.

Number of features	Model	c-index	Log rank test statistic	Hazard ratio
10	SVR	0.562 ± 0.015	12.281 ± 4.955	1.240 ± 0.063
	Cox	0.560 ± 0.013	10.662 ± 5.234	1.245 ± 0.060
20	SVR	0.552 ± 0.013	8.589 ± 5.557	1.203 ± 0.048
	Cox	0.552 ± 0.013	9.289 ± 5.990	1.194 ± 0.054
30	SVR	0.540 ± 0.010	5.070 ± 2.275	1.156 ± 0.042
	Cox	0.541 ± 0.011	5.510 ± 3.907	1.154 ± 0.041
40	SVR	0.535 ± 0.011	4.900 ± 2.907	1.117 ± 0.029
	Cox	0.538 ± 0.011	4.367 ± 2.573	1.125 ± 0.035
50	SVR	0.540 ± 0.014	5.234 ± 3.557	1.136 ± 0.054
	Cox	0.537 ± 0.013	3.874 ± 2.681	1.120 ± 0.047
60	SVR	0.532 ± 0.012	3.474 ± 2.748	1.105 ± 0.039
	Cox	0.528 ± 0.015	3.723 ± 2.662	1.112 ± 0.052
70	SVR	0.535 ± 0.012	4.134 ± 2.304	1.123 ± 0.036
	Cox	0.528 ± 0.016	2.774 ± 1.925	1.096 ± 0.044
80	SVR	0.530 ± 0.010	3.171 ± 2.597	1.120 ± 0.036
	Cox	0.526 ± 0.010	1.687 ± 1.323	1.079 ± 0.041
90	SVR	0.526 ± 0.013	2.846 ± 2.219	1.098 ± 0.030
	Cox	0.521 ± 0.012	1.720 ± 1.492	1.070 ± 0.041
100	SVR	0.524 ± 0.012	1.594 ± 1.386	1.085 ± 0.045
	Cox	0.517 ± 0.014	1.226 ± 1.033	1.054 ± 0.030
110	SVR	0.535 ± 0.012	3.738 ± 3.335	1.101 ± 0.048
	Cox	0.515 ± 0.008	0.870 ± 0.800	1.057 ± 0.024
120	SVR	0.527 ± 0.012	1.863 ± 1.581	1.091 ± 0.035
	Cox	0.515 ± 0.015	1.092 ± 0.964	1.058 ± 0.033

value < 0.05, value < 0.01, value < 0.001 (Wilcoxon rank sum test).