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

Table 1 Performance of the prediction model with different types of features in the fivefold cross validation

Cancer	Feature	#Features	#PCs	SN	SP	ACC	PPV	NPV	AUC
BRCA	EXP	5119	430	0.674	0.709	0.692	0.694	0.689	0.691
BRCA	\(\Delta\)PCC	1563	480	0.773	0.806	0.790	0.796	0.784	0.789
COAD	EXP	835	100	0.360	0.935	0.758	0.711	0.767	0.647
COAD	\(\Delta\)PCC	1969	80	0.760	0.965	0.902	0.905	0.901	0.862
HNSC	EXP	292	10	0.750	0.684	0.720	0.739	0.696	0.717
HNSC	\(\Delta\)PCC	800	100	0.956	0.877	0.920	0.903	0.943	0.917
LUAD	EXP	6193	110	0.477	0.882	0.741	0.683	0.759	0.679
LUAD	\(\Delta\)PCC	12,981	200	0.593	0.944	0.822	0.850	0.813	0.769
LUSC	EXP	1371	190	0.644	0.867	0.786	0.736	0.809	0.756
LUSC	\(\Delta\)PCC	2436	200	0.875	0.934	0.912	0.884	0.929	0.904
STAD	EXP	476	120	0.905	0.472	0.763	0.778	0.708	0.688
STAD	\(\Delta\)PCC	17,445	60	0.973	0.903	0.950	0.953	0.942	0.938
THCA	EXP	4205	30	0.663	0.663	0.663	0.634	0.691	0.663
THCA	\(\Delta\)PCC	3397	150	0.674	0.723	0.700	0.682	0.716	0.698

In comparison of two types of features (RNA expression vs. deltaPCC), the better performances are shown in bold
In all cancer types, prediction with \(\Delta\)PCCs showed a better performance than that with RNA expression levels
PC, principal component; SN, sensitivity; SP, specificity; ACC, accuracy; PPV, positive predictive value; NPV, negative predictive value; AUC, area under the curve; EXP, RNA expression level

ISSN: 1755-8794