Diagnosis of Alzheimer’s Disease with Extreme Learning Machine on Whole-Brain Functional Connectivity

Table 3

Comparison of classification performances with references.

Reference	Data set AD : MCI:CN	Feature measures	Feature selection	Classifier	Accuracy (%)		AUC
					AD vs. CN	MCI vs. CN	AD vs. CN	MCI vs. CN
[14]	-:25 : 25	Pearson correlation of regional cortical thickness	t-test; mRMR; SVM-RFE;	SVM	92.35	84	0.9744	0.9233
[15]	-:12 : 25	FC	t-test; SVM-RFE;	SVM	N/A	91.9	N/A	0.94
[27]	25:-:36	FC	Random neural network cluster	Elman neural network	92.31	N/A	N/A	N/A
[17]	34 : 31 : 31	ReHo; FC	SVM-RFE; LASSO; t-test	ELM	98.86	98.57	N/A	N/A
[18]	118 : 118 : 118	FC	Recurrent learning method; convolutional learning method;	ELM	N/A	N/A	0.913	0.824
[19]	118 : 118 : 118	FC	Select features by threshold	GNEA	N/A	N/A	0.813	0.703
[20]	31 : 31 : 31	FC, graph-embedding	SVM-RFE; LASSO; FSASL	LSVM	90.63	97.8	N/A	N/A
				RELM	93.86	98.91	N/A	N/A
Proposed method	100:100:100	FC	None	Parallel ELM	96.85	95.05	0.9891	0.9888