genomemedicine.com

Table 2
LS-SVM models for the prediction of WHEELER, pN-STAGE and CRM in rectal cancer
Outcome	Model	NG*	NP^†	AUC (SE)^‡	p-value^§

WHEELER
A	MT₀	4		0.7538 (0.1085)	0.0987
	MT₁	29		0.9038 (0.0502)	0.6861
	PT₀		35	0.7423 (0.0867)	0.0540
	PT₁		11	0.9038 (0.0575)	0.7273

B	MT₀-T₁	32		0.6846 (0.1215)	0.0598
	PT₀-T₁		5	0.8654 (0.0621)	0.4135

C	MT₀₁	3^¶		0.7808 (0.0985)	0.1320
	PT₀₁		21^¶	0.7692 (0.0831)	0.0831
	MPT₀	3	35	0.8461 (0.0718)	0.2760
	MPT₁	25	12	0.9269 (0.0425)
	MPT₀₁	2^¶	31^¶	0.8846 (0.0558)	0.4858

	MT₀PT₁	2	4	0.9385 (0.0444)	0.8101^¥

pN-STAGE
A	MT₀	25		0.6493 (0.0914)	2.315e-4
	MT₁	22		0.8506 (0.0665)	0.0362
	PT₀		2	0.6753 (0.0906)	6.659e-4
	PT₁		12	0.8409 (0.0652)	0.0238

B	MT₀-T₁	4		0.6071 (0.0986)	1.359e-4
	PT₀-T₁		9	0.7662 (0.0900)	0.0153

C	MT₀₁	24^¶		0.9286 (0.0450)	0.1998
	PT₀₁		34^¶	0.8182 (0.0695)	0.0145
	MPT₀	27	27	0.9188 (0.0469)	0.1591
	MPT₁	21	14	0.9870 (0.0135)
	MPT₀₁	23^¶	16^¶	0.9610 (0.0280)	0.3421

	MT₀PT₀₁	26	20^¶	1 (0)	0.3347^¥

CRM
A	MT₀	33		0.6790 (0.1016)	0.0072
	MT₁	9		0.9259 (0.0472)	0.4955
	PT₀		34	0.8518 (0.0624)	0.0935
	PT₁		34	0.7654 (0.0831)	0.0281

B	MT₀-T₁	6		0.9136 (0.0480)	0.4030
	PT₀-T₁		2	0.8272 (0.0709)	0.0849

C	MT₀₁	16^¶		0.8066 (0.0846)	0.0468
	PT₀₁		3^¶	0.7531 (0.0865)	0.0227
	MPT₀	7	27	0.8477 (0.0688)	0.1340
	MPT₁	7	33	0.9630 (0.0344)
	MPT₀₁	2^¶	3^¶	0.8230 (0.0771)	0.0973

	MT₁PT₀	16	14	0.9630 (0.0376)	1
	MT₀₁PT₁	9^¶	29	0.9876 (0.0146)	0.4924^¥

*Number of genes selected in each LOO iteration. ^†Number of proteins selected in each LOO iteration. ^‡Area under the ROC curve (standard error) obtained with leave-one-out. ^§Comparison of AUC between each model and the best model in bold [46]. ^¶Number of features used at both time points. ^¥This model is better than the model in bold we compare with.
Daemen et al. Genome Medicine 2009 1:39 doi:10.1186/gm39

Table 2