Table 3.

A summary (mean of 100 iterations) of the classification accuracy and AUC using the Enet and proposed Enet-subset feature selection methods.

Biomarker(s)	Using all Enet selected features		Using Enet-subset selected features
	ACC (%), AUC (mean)	Features (mean/total #)	ACC (%), AUC (mean)	Features (mean/total #)
BC	81.7, 0.919	349/360	82.6, 0.920	326/360
CC	81.0, 0.92	349/360	82.3, 0.925	328/360
DC	80.9, 0.898	348/360	81.2, 0.895	324/360
LE	50.8, 0.598	348/360	50.4, 0.590	155/360
BC+CC	81.0, 0.923	679/720	82.5, 0.92	634/720
BC+DC	81.2, 0.907	680/720	83.2, 0.924	636/720
CC+DC	80.8, 0.913	680/720	81.8, 0.921	640/720
BC+CC+DC	80.9, 0.916	1,006/1,080	83.1, 0.937	945/1,080

ACC, Accuracy; AUC, Area under curve; BC, Betweenness centrality; CC, Clustering coefficient; DC, Degree centrality; LE, Local efficiency.