A unified frame of predicting side effects of drugs by using linear neighborhood similarity

Table 5 Performances of LNSM-MSE and benchmark methods evaluated by 5-CV

Dataset	Methods	AUPR	AUC	SN	SP	Precision	Accuracy	F
Pauwels’s dataset	Liu’s method	0.345	0.920	0.643	0.950	0.400	0.934	0.493
	Cheng’s method	0.588	0.922	0.587	0.975	0.547	0.955	0.566
	RBMBM	0.612	0.941	0.605	0.977	0.579	0.958	0.592
	INBM	0.641	0.934	0.608	0.979	0.605	0.961	0.607
	LNSM-MSE	0.671	0.948	0.629	0.980	0.625	0.963	0.627
Mizutani’s dataset	Liu’s method	0.366	0.918	0.637	0.948	0.418	0.930	0.505
	Cheng’s method	0.599	0.923	0.593	0.973	0.560	0.951	0.576
	RBMBM	0.619	0.939	0.614	0.974	0.581	0.954	0.597
	INBM	0.646	0.932	0.616	0.976	0.605	0.956	0.611
	LNSM-MSE	0.676	0.944	0.627	0.979	0.635	0.959	0.631
Liu’s dataset	Liu’s method	0.278	0.907	0.669	0.930	0.341	0.917	0.452
	Cheng’s method	0.592	0.922	0.589	0.974	0.550	0.954	0.569
	RBMBM	0.616	0.941	0.608	0.976	0.581	0.957	0.594
	INBM	0.641	0.934	0.607	0.979	0.606	0.959	0.606
	LNSM-MSE	0.673	0.948	0.631	0.979	0.624	0.962	0.628