TABLE 3.

Feature coding methods based on 5-fold cross-validation of the training dataset (benchmark dataset 1).

Encoding	Sn	SP	Acc	MCC	AUC
One-hot	0.6632 ± 0.0032	0.9591 ± 0.0026	0.9358 ± 0.0042	0.5715 ± 0.0015	0.9644 ± 0.0011
NCP	0.6904 ± 0.0035	0.9568 ± 0.0064	0.9360 ± 0.0028	0.5959 ± 0.0030	0.9645 ± 0.0014
One-hot + NCP	0.7510 ± 0.0016	0.9512 ± 0.0045	0.9356 ± 0.0032	0.6584 ± 0.0021	0.9666 ± 0.0009

The highest value for each evaluation metric is highlighted in bold.