Table 5.

Performance comparison of different feature sets with lasso-based feature selection in AD vs. MCI classification

	AD vs. MCI
	LLF	SAEF	LLF + SAEF
Supervised
MRI	0.617 ± 0.020	0.631 ± 0.023	0.704 ± 0.026
PET	0.667 ± 0.023	0.645 ± 0.015	0.711 ± 0.025
CSF	0.659 ± 0.004	0.661 ± 0.002	0.655 ± 0.009
CONCAT	0.693 ± 0.019	0.681 ± 0.023	0.752 ± 0.030
MK-SVM	0.788 ± 0.018	0.759 ± 0.019	0.827 ± 0.025
Semi-supervised
MRI	–	0.659 ± 0.025	0.721 ± 0.039
PET	–	0.640 ± 0.021	0.715 ± 0.024
CSF	–	0.659 ± 0.002	0.659 ± 0.005
CONCAT	–	0.682 ± 0.022	0.781 ± 0.028
MK-SVM	–	0.757 ± 0.017	0.837 ± 0.015

Bold best performance across both the feature types and the learning schemes, italics best performance across the feature types in the same learning scheme