Skip to main content

Table 5 GBM subtype prediction.

From: Evaluation of data discretization methods to derive platform independent isoform expression signatures for multi-class tumor subtyping

Class PN N CL M
Method Sn Sp Sn Sp Sn Sp Sn Sp
SVM 0.681 0.963 0.818 1.000 0.772 0.926 0.863 0.870
RF 0.833 0.965 0.944 1.000 0.888 0.965 0.833 0.982
NB 0.950 0.946 1.000 0.910 0.850 0.928 0.800 0.910
PAM 0.937 0.916 0.875 0.966 0.875 0.966 0.750 0.983
  1. Sensitivity (Sn) and Specificity (Sp) of the classifiers trained on the top 100 feature variables from exon-array data and tested on the independent RNA-seq data for prediction of the four GBM Subtypes.