. Author manuscript; available in PMC: 2017 Jun 1.

Published in final edited form as: Genomics. 2016 Apr 30;107(6):223–230. doi: 10.1016/j.ygeno.2016.04.005

Table 2.

Prediction performance of different combinations of clinical and omics measurements: mean (sd) of C-statistic.

Model	Enet	SPCA	SPLS
Cln	0.708 (0.077)	0.610 (0.076)	0.714 (0.023)
Cln+Mu	0.707 (0.078)	0.612 (0.072)	0.710 (0.021)
Gen	0.575 (0.071)	0.662 (0.063)	0.665 (0.022)
Met	0.605 (0.068)	0.668 (0.067)	0.702 (0.021)
CAN	0.501 (0.069)	0.570 (0.071)	0.586 (0.026)
Cln+Mu+Gen	0.705 (0.076)	0.661 (0.072)	0.713 (0.022)
Cln+Mu+Met	0.724 (0.071)	0.707 (0.060)	0.743 (0.020)
Cln+Mu+CNA	0.691 (0.083)	0.626 (0.069)	0.722 (0.023)
Cln+Mu+Gen+Met	0.717 (0.070)	0.714 (0.059)	0.746 (0.021)
Cln+Mu+Gen+CNA	0.686 (0.076)	0.660 (0.073)	0.714 (0.022)
Cln+Mu+Met+CNA	0.713 (0.073)	0.718 (0.061)	0.743 (0.020)
Cln+Mu+Gen+Meth+CNA	0.706 (0.072)	0.707 (0.059)	0.746 (0.021)

Cln: clinical variables; Mu: mutation; Gen: mRNA gene expression; Met: methylation; CNA: copy number alternation. A larger value of C-statistic indicates better prediction. A value of 0.5 corresponds to a model with no predictive power.