. 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 V.

Performance Comparison Based on the Optimal Number of Features Selected in the TU Cohort

TU Cohort	Metabolomics	Integrated (P + G + M)
Accuracy	0.86	0.90
Sensitivity	0.91	0.91
Specificity	0.84	0.89
AUC (95% CI)	0.93 (0.84, 0.97)	0.99 (0.95, 0.99)
Optimal # of Features	14/50	21/233
Selected Features^b	L-glutamic acid^a L-valine^a L-(+) lactic acid^a N-acetyl-5-hydroxytryptamine L-threonine Diglycerol Urea Arachidic acid Trans-aconitic acid L-proline N, N-dimethyl-1 4-phenylenediamine D-glucose L-serine L-cystine	P01024 ^a P01591 P02743 ^a P02763 P05160 ^a P06727 P13591 P13598 ^a P22891 ^a P35858 [43000] [53000]^a [63423] [28000] [66012]^a L-glutamic acid^a L-valine^a L-(+) lactic acid^a L-threonine Urea L-cystine

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

The results of best performing methods are marked in bold.