Classifying breast cancer subtypes on multi-omics data via sparse canonical correlation analysis and deep learning

Table 6 Performance of different DSCCN models on each binary classification dataset

Breast Cancer Subtype dataset	DSCCN1	DSCCN2	DSCCN3	DSCCN4	DSCCN5	DSCCN
Accuracy
Basal vs Her2	0.833	0.867	0.880	0.924	0.842	0.926
Basal vs LumA	0.947	0.958	0.958	0.978	0.767	0.982
Basal vs LumB	0.943	0.953	0.948	0.953	0.610	0.965
Her2 vs LumA	0.905	0.937	0.914	0.936	0.905	0.951
Her2 vs LumB	0.896	0.902	0.913	0.843	0.825	0.926
LumA vs LumB	0.695	0.768	0.766	0.786	0.635	0.844
AUC
Basal vs Her2	0.950	0.960	0.984	0.973	0.957	0.982
Basal vs LumA	0.972	0.990	0.980	0.997	0.979	0.997
Basal vs LumB	0.965	0.963	0.997	0.933	0.965	0.997
Her2 vs LumA	0.946	0.970	0.969	0.966	0.981	0.948
Her2 vs LumB	0.910	0.949	0.934	0.943	0.650	0.951
LumA vs LumB	0.857	0.808	0.878	0.847	0.757	0.857
F1-score
Basal vs Her2	0.742	0.769	0.833	0.822	0.686	0.933
Basal vs LumA	0.965	0.975	0.967	0.986	0.848	0.988
Basal vs LumB	0.962	0.968	0.943	0.972	0.694	0.974
Her2 vs LumA	0.950	0.966	0.955	0.962	0.950	0.973
Her2 vs LumB	0.906	0.940	0.896	0.909	0.902	0.956
LumA vs LumB	0.667	0.642	0.695	0.598	0.401	0.883

ISSN: 1471-2105