Considerations for feature selection using gene pairs and applications in large-scale dataset integration, novel oncogene discovery, and interpretable cancer screening

Table 7 Performance of feature selection methods and ensemble classifiers used to build STELs from circulating miRNA pairs

Cancer	Feature Selection	Model	Accuracy	Pairs	STEL Accuracy
Bladder	TSP	RF	0.643	83	0.622
	AM	RF	0.622	97	0.556
	AA	RF	0.628	91	0.577
	FM	RF	0.628	87	0.566
	FA	RF	0.638	94	0.577
	TSP	BT	0.633	99	0.602
	AM	BT	0.612	82	0.536
	AA	BT	0.622	96	0.531
	FM	BT	0.617	92	0.536
	FA	BT	0.617	97	0.541
Ovary	TSP	RF	0.901	90	0.896
	AM	RF	0.909	82	0.896
	AA	RF	0.905	59	0.896
	FM	RF	0.958	77	0.936
	FA	RF	0.956	87	0.927
	TSP	BT	0.899	96	0.896
	AM	BT	0.906	96	0.896
	AA	BT	0.904	78	0.896
	FM	BT	0.957	66	0.896
	FA	BT	0.960	98	0.896
Pancreas	TSP	RF	0.832	93	0.772
	AM	RF	0.712	81	0.688
	AA	RF	0.696	23	0.688
	FM	RF	0.868	69	0.848
	FA	RF	0.864	80	0.828
	TSP	BT	0.796	93	0.744
	AM	BT	0.708	39	0.688
	AA	BT	0.700	97	0.688
	FM	BT	0.900	97	0.772
	FA	BT	0.872	96	0.748
Prostate	TSP	RF	0.762	83	0.740
	AM	RF	0.780	80	0.766
	AA	RF	0.755	100	0.740
	FM	RF	0.747	8	0.740
	FA	RF	0.747	8	0.740
	TSP	BT	0.762	94	0.740
	AM	BT	0.762	36	0.740
	AA	BT	0.740	1	0.740
	FM	BT	0.740	1	0.740
	FA	BT	0.740	1	0.740

ISSN: 1755-8794