Table 2.

The performance for the AD vs. pMCI vs sMCI task.

		Testing performance				Cross-validation performance
Model	Dataset	A	P	R	F1	A	P	R	F1
RF	CS	87.76	87.76	87.79	87.76	90.35 ± 2.9	90.49 ± 2.8	90.47 ± 2.7	90.28 ± 2.5
	AM	68.05	67.47	68.82	68.05	58.39 ± 4.1	58.3 ± 4.26	60.74 ± 4.3	58.27 ± 4.2
	AM-CS	88.54	88.51	88.92	88.54	92.74 ± 3.1	92.87 ± 3.1	93.03 ± 2.6	93.21 ± 3.3
DT	CS	90.89	90.94	91.15	90.89	83.08 ± 4.9	83.08 ± 4.9	83.49 ± 4.9	83.13 ± 4.8
	AM	64.81	64.64	69.15	64.81	53.23 ± 4.3	53.23 ± 4.6	55.27 ± 4.7	53.21 ± 4.2
	AM-CS	89.32	89.30	89.74	89.32	84.96 ± 3.7	84.95 ± 3.7	85.61 ± 3.5	85.15 ± 3.7
LR	CS	76.85	77.15	78.34	76.85	79.85 ± 4.4	79.96 ± 4.3	80.61 ± 4.2	79.85 ± 4.4
	AM	53.24	51.32	52.76	53.24	55.39 ± 4.6	54.64 ± 4.6	54.7 ± 4.71	55.39 ± 4.6
	AM-CS	85.42	85.40	85.64	85.42	88.40 ± 3.3	88.36 ± 3.3	88.91 ± 3.1	88.4 ± 3.28
SVM	CS	81.02	81.11	81.40	81.02	79.95 ± 3.9	79.93 ± 4.0	80.7 ± 3.96	79.95 ± 3.9
	AM	58.33	57.05	58.85	58.33	55.35 ± 4.5	55.52 ± 4.4	56.54 ± 4.3	55.35 ± 4.5
	AM-CS	76.39	76.55	77.04	76.39	82.89 ± 3.1	82.78 ± 3.1	83.18 ± 3.1	82.89 ± 3.1
KNN	CS	73.61	74.31	76.53	73.61	76.32 ± 3.7	76.8 ± 3.68	80.52 ± 3.4	76.32 ± 3.7
	AM	48.00	42.67	45.20	48.00	52.31 ± 3.2	43.13 ± 3.4	39.62 ± 7.5	52.31 ± 3.2
	AM-CS	74.22	72.37	82.99	74.22	82.43 ± 4.2	82.11 ± 4.4	84.62 ± 3.8	82.43 ± 4.2