Table 4.

Comparison the LASSO and Ridge regression methods with Elastic Net regression

		Univariate		Multivariate
		HR (95 % CI)	P	HR (95 % CI)	P
Datasets	Methods
	Lasso	0.106 (0.030 – 0.370)	0.000	0.063 (0.016 – 0.252)	0.000
Our dataset	Ridge	0.812 (0.303 – 2.173)	0.000	0.083 (0.022 – 0.321)	0.000
	Elastic net	0.065 (0.014 – 0.287)	0.000	0.040 (0.008 – 0.198)	0.000
Denmark dataset	Lasso	0.055 (0.007 – 0.453)	0.007	0.044 (0.005 – 0.396)	0.005
	Ridge	0.112 (0.024 – 0.519)	0.005	0.080 (0.014 – 0.467)	0.005
	Elastic net	0.057 (0.007 – 0.464)	0.007	0.038 (0.004 – 0.389)	0.005
Australian dataset	Lasso	0.565 (0.396 – 0.805)	0.002	0.549 (0.384 – 0.784)	0.001
	Ridge	0.447 (0.312 – 0.641)	0.000	0.454 (0.316 – 0.651)	0.000
	Elastic net	0.523 (0.370 – 0.739)	0.000	0.529 (0.373 – 0.748)	0.000
USA dataset	Lasso	0.105 (0.010 – 1.068)	0.056	0.104 (0.010 – 1.052)	0.055
	Ridge	0.130 (0.013 – 1.294)	0.082	0.129 (0.013 – 1.283)	0.081
	Elastic net	0.120 (0. 012 – 1.195)	0.071	0. 122 (0. 012 – 1.214)	0. 072
Norway dataset	Lasso	---	---	---	---
	Ridge	---	---	---	---
	Elastic net	0.536 (0.300 – 0.957)	0.035	0.569 (0.318 – 1.018)	0.057

This table shows the comparison with the LASSO, Ridge regression and Elastic Net methods for 19 gene signatures based on our dataset and other external datasets from different countries. Univariate and multivariate Cox’s proportional hazard model analysis of prognostic factor (prognostic index or risk score) for overall survival

[Relevant location: Page 16]