Table 5.

The predictive performance of the different methods in MCI conversion study.

Reference	Method	Conversion time (month)	sMCI (n)	pMCI (n)	Accuracy (%)	Sensitivity (%)	Specificity (%)	AUC
Young et al.	Gaussian process	36	96	47	65.0	66.0	64.6	0.767
Liu et al.	Independent component analysis and Cox model	36	108	126	68.8	57.1	82.4	0.736
Lange et al.	SPM t-test	36	77	31	/	/	/	0.832
Kengo et al.	Logistic regression	24	47	41	83	70	90	/
Lu et al.	Deep neural network	36	409	217	81.5	78.2	82.5	/
Pagani et al.	Independent components analysis	60	27	95	83.5	83.2	85.2	0.894
Proposed	Metabolic CPM	36	242	178	85.2	88.1	86.4	0.933