Figure 3.

SVM modeling and classification accuracy evaluation. (A) Accurate rate using crucial genes (screened by the betweenness centrality score) as classifier; 100% accuracy was achieved in the top 40. (B) Visualization of classification result using the 16-gene signature as an SVM classifier and GSE68793 as a training dataset (orange, stage II sample; blue, stage I sample). (C) Hierarchical clustering analysis indicated samples of the same type, which were clustered using the 16-gene signature (yellow, higher expression; blue, lower expression). SVM, support vector machine.