. Author manuscript; available in PMC: 2011 Sep 27.

Published in final edited form as: Pharmacogenomics J. 2010 Aug;10(4):267–277. doi: 10.1038/tpj.2010.33

Table 5.

Prediction accuracies of gene- and pathway-based classifiers for independent (validation) data set samples.

	Classifier	Independent data sets			Mean Acc
	Classifier	Acetaminophen	Carbon Tetrachloride	Allyl Alcohol	Mean Acc
Pathway- based classifiers^*	RF-a	0.855	0.875	0.642	0.791
	RF-b	0.888	0.903	0.642	0.811
	RF-c	0.882	0.889	0.705	0.825
Gene-based classifiers^**	RF	0.816	0.889	0.684	0.796
	KNN	0.836	0.889	0.684	0.803
	SVM	0.888	0.931	0.747	0.855
	NC	0.921	0.917	0.747	0.862
	Mean Acc	0.869	0.899	0.693	0.820

Pathway-based random forest (RF) classifiers (Table 2) consist of genes in the a) regulation of apoptosis by mitochondrial proteins pathway, b) Anti-apoptotic TNFs/NF-kB/Bcl-2 pathway, or c) Toll-like receptor (TLR) ligands and common TLR signaling pathway.

^**

Gene-based pathways are developed using the entire blood data set.