Discriminating lymphomas and reactive lymphadenopathy in lymph node biopsies by gene expression profiling

BMC Medical Genomics

Table 2 The accuracy rates resulting from GEP classification of lymph node biopsies into selected subtypes.

Comparison	Subtypes	Optimal number of probes	Training set accuracy rate (%)	Test set accuracy rate* (%)
Global multi-class	RL v cHL v FL v DLBCL	50	83.6	68.6
Local binary-class	RL v Lymphoma	130	87.7	80.0
	cHL v NHL	40	89.2	82.1
	FL v DLBCL	10	84.8	76.1
Global binary-class	cHL v remaining cases	30	91.4	82.8
	FL v remaining cases	60	82.7	88.5
	DLBCL v remaining cases	490	87.7	82.8

* The reported independent test set accuracy rates indicate the proportion of correctly classified cases from the total number assessed in the test datasets.

ISSN: 1755-8794