. Author manuscript; available in PMC: 2017 Sep 1.

Published in final edited form as: IEEE J Biomed Health Inform. 2016 May 27;20(5):1225–1231. doi: 10.1109/JBHI.2016.2574201

TABLE 6.

Performance Comparison Based on the Optimal Number of Features Selected in the GU Cohort (44 Samples)

GU Cohort	Proteomics	Glycomics		Metabolomics	Integrated (P + G)		Integrated (P+G+M)
Accuracy	0.89	0.91		0.84	0.98		0.98
Sensitivity	0.94	0.87		0.85	0.95		0.96
Specificity	0.85	0.95		0.83	0.99		0.99
AUC (95% CI)	0.87 (0.72, 0.96)	0.97 (0.89, 0.99)		0.91 (0.77, 0.97)	0.99 (0.95, 0.99)		0.99 (0.95, 0.99)
Optimal Number of Features	3/101	10/82		4/50	15/183		10/233
Selected Features^b	O75636 ^a P00736 P00751 ^a	[43100]^a		Ethanolamine L-(+) lactic acid^a Oxalic acid Putrescine	O75636 ^a P01023 ^a P02774 ^a P04278 P16070 P41222 ^a P80108 ^a [53111]^c	[53313] [34110] [43110]^a [43200]^a [43201] [73514] [53111]^a,^c	O75636 ^a P01876 ^a P14151 P41222 ^a [43100]^a [53101]^a	[53111]^a Malonic acid Putrescine Sorbose^a
		[53313]	[53411]^a
		[53000]^a	[43201]
		[34110]	[63402]
		[53100]	[53111]^a
		[53302]

Significant (p value ≤ 0.05) in univariate statistical analysis

N-glycans are characterized by GlcNAc, mannose, galactose, fucose, and NeuNAc, and proteins are indicated by Uniprot IDs

Isomers with different retention times.

The results of best performing methods are marked in bold.