Identification of recent cases of hepatitis C virus infection using physical-chemical properties of hypervariable region 1 and a radial basis function neural network classifier

Table 2 RBFNN performance in R/C classification of Intra-host HVR1 PhyChem variants^a

Dataset	CA	F₁ measure	MCC	AUROC
Full train set^b	95.795%	0.958	0.910	0.986
Train set	94.847%^c	0.948^c	0.890^c	0.979^c
Test set	84.145%^d	0.842^d	0.670^d	0.912^d
Random-labeled train set	59.038%^e (±1.28)	0.521^e (±0.007)	−0.007^e (±0.022)	0.501^e (±0.012)
Test set	39.965%^f (±1.948)	0.280^f (±0.070)	0.003^f (±0.145)	0.385^f (±0.144)

^aFor description of train/test data, see Methods Section
^bValues obtained from RBFNN classifier trained on entire training dataset without CV
^cOverall value represents averaged values of 10xCV data
^dValue obtained from RBFNN classifier trained on training dataset by 10xCV
^eOverall value represents averaged values of 10xCV data obtained from 4 datasets. Standard deviation (SD), in parenthesis
^fOverall value represents averaged values obtained from 4 RBFNN classifiers trained on randomly-labeled data by 10xCV (SD)

ISSN: 1471-2164