Accurate breast cancer diagnosis using a stable feature ranking algorithm

Table 7 Representative achievement on the BC datasets

		n	FR/SFS	Classifier	AUC	ACC	SEN	SPE
BCDR-F03	[19]	600		SVM	0.77±0.03
	[19]	17		SVM	0.77±0.02
	[53]	4	elastic net	SVM	0.69±0.05	0.74±0.05	0.56±0.10	0.81±0.08
	Ours	4	GFS	NB	0.71±0.04	0.77±0.03	0.59±0.10	0.84±0.05
WDBC	[54]	24	variable importance	hierarchical clustering RF	0.9896	0.9705	0.9477	0.9841
	[55]	14	genetic algorithm	particle swarm optimization		0.966	0.975	0.937
	[48]	6	genetic algorithm	kernel-based Bayesian	0.994	0.971	0.924	1.000
	[56]	14	genetic algorithm	rotation forest	0.993	0.9948
	[57]	9	interaction dominance			0.9966
	Ours	2	GFS	NB	0.94±0.02	0.94±0.01	0.94±0.03	0.94±0.02
GSE10810	[22]	8088	false discovery rate			1.000
	[49]	80	t-test	SVM	0.7789
	Ours	2	GFS	SVM	0.96±0.05	0.97±0.04	0.99±0.03	0.92±0.10
GSE15852	[23]	33	paired t-test	hierarchical cluster analysis		0.88	0.86	0.91
	[51]	10	logistic regression	RF		0.9311
	[52]	50	prioritization analysis	SVM		0.87
	Ours	4	GFS	NB	0.89±0.07	0.89±0.07	0.96±0.07	0.81±0.13

ISSN: 1472-6947