Table 5.

A summary of the performances for pMCI vs. sMCI classification

Method	Modality	ACC (%)	SEN (%)	SPEC (%)	BAC (%)	PPV (%)	NPV (%)
SMTL	MRI	51.44 ± 3.68	39.50	61.00	50.25	44.78	55.74
	PET	50.92 ± 3.97	40.50	59.33	49.92	44.56	55.27
	MP	54.48 ± 3.88	41.50	65.00	53.25	49.01	57.82
	MPC	60.69 ± 4.06	46.50	72.00	59.25	56.96	62.80
DW-SMTL	MRI	55.46 ± 3.65	29.00	76.33	52.67	49.14	57.68
	PET	54.12 ± 4.28	38.00	66.33	52.17	46.09	58.55
	MP	56.71 ± 4.28	44.00	67.00	55.50	51.91	59.64
	MPC	58.56 ± 4.09	36.50	75.67	56.08	53.78	60.57
S2MTL	MRI	61.00 ± 3.47	53.00	66.33	59.67	51.19	67.93
	PET	57.87 ± 4.13	44.50	68.33	56.42	52.36	61.15
	MP	65.33 ± 3.84	59.50	70.67	65.08	65.01	65.58
	MPC	67.04 ± 4.38	67.00	66.67	66.83	61.48	69.89
D-S2MTL	MRI	63.71 ± 2.43	25.50	92.00	58.75	70.24	62.52
	PET	55.25 ± 4.60	50.00	59.67	54.83	51.07	58.64
	MP	67.82 ± 3.41	37.00	93.33	65.17	82.12	64.16
	MPC	70.73 ± 3.41	42.50	93.00	67.75	82.73	67.22
DW-S2MTL	MRI	69.84 ± 2.68	44.00	89.00	66.50	74.79	68.19
	PET	65.71 ± 3.64	29.50	95.00	62.25	82.68	62.49
	MP	74.15 ± 3.35	50.50	92.67	71.58	84.36	70.51
	MPC	73.04 ± 3.51	53.00	89.00	71.00	79.33	70.39

Boldface denotes the best performance and the maximum performance in each metric

SMTL sparse multi-task learning, S ²MTL subclass-based SMTL, DW-SMTL deep weighted SMTL, D-S ²MTL deep S²MTL, DW-S ²MTL deep weighted S²MTL