Table 1.

Comparison between the dimension reduction methods on the clinical data

Classifier	Dimension reduction method		Number of clinical labels /components	FPR	TPR	F0.5 score	Kappa
Gaussian Naïve Bays	N/A	Raw data	67 labels	0.33	0.85	0.71	0.45
	Clinical labels selection	SelectKBest	13 labels	0.47	0.89	0.71	0.41
	Clinical labels selection	ExtraTree	30 labels	0.37	0.89	0.75	0.51
	Feature extraction	PCA	25 components	0.51	0.94	0.74	0.46