TABLE 5.

Feature coding methods based on 5-fold cross-validation of the training dataset (benchmark dataset 2).

Encoding	Sn	SP	Acc	MCC	AUC
One-hot	0.8707 ± 0.0053	0.9771 ± 0.0049	0.9635 ± 0.0028	0.8381 ± 0.0019	0.9883 ± 0.0001
NCP	0.8567 ± 0.0037	0.9797 ± 0.0035	0.9639 ± 0.0016	0.8380 ± 0.0035	0.9883 ± 0.0004
One-hot + NCP	0.8748 ± 0.0036	0.9788 ± 0.0042	0.9655 ± 0.0011	0.8466 ± 0.0025	0.9889 ± 0.0001

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