Predicting lymph node metastasis and prognosis of individual cancer patients based on miRNA-mediated RNA interactions

Table 3 Comparison of the performance of our SVM model with that of Zhang’s SVM model [8]

Cancer	Method_feature	Train_score	Test_score
BRCA	Our model_\(\Delta\)PCC	0.972	0.787
	Zhang_mRNA	0.798	0.680
	Zhang_miRNA	0.764	0.737
	Zhang_lncRNA	0.793	0.696
COAD	Our model_\(\Delta\)PCC	0.984	0.905
	Zhang_mRNA	0.849	0.871
	Zhang_miRNA	0.902	0.886
	Zhang_lncRNA	0.869	0.871
LUAD	Our model_\(\Delta\)PCC	0.996	0.850
	Zhang_mRNA	0.808	0.849
	Zhang_miRNA	0.885	0.795
	Zhang_lncRNA	0.798	0.849
LUSC	Our model_\(\Delta\)PCC	0.999	0.904
	Zhang_mRNA	0.871	0.900
	Zhang_miRNA	0.939	0.847
	Zhang_lncRNA	0.861	0.900

In comparison of two types of features (RNA expression vs. deltaPCC), the better performances are shown in bold
Among the seven types of cancer used in our study, comparison was made in four types of cancer because they are the only common cancer types in both studies. The train_score and test_score were obtained using the scikit-learn package, which was used by Zhang’s study. In all four caner types, our model showed the better performance in both training and testing. our model_\(\Delta\)PCC: SVM model using \(\Delta\)PCCs as features. Zhang_X: SVM model using the expression levels of RNA type X as features

ISSN: 1755-8794