TABLE 10.

Comparison of the performance of different model algorithms in experiments for LMCI vs. AD classification with the related works.

Method	Modality	DATA (LMCI/AD)	ACC	SEN	SPE	AUC
SVM	PET	147/198	0.5841	0.7834	0.5044	0.6908
Hao et al., 2020	PET	273/187	0.6677	0.7545	0.5594	0.68
	MRI		0.712	0.7801	0.6332	0.76
Lei et al., 2020	fMRI	44/38	0.7805	0.7368	0.8182	0.8571
	DTI		0.5366	0.5789	0.5000	0.5260
Song et al., 2021	fMRI	44/38	0.7926	0.8421	0.75	0.9067
	DTI		0.8292	0.9473	0.7272	0.9414
Our method (Tri-loss+BP)	PET	147/198	0.8118	0.7753	0.8441	0.8167

The bold values represent the highest number.