Table 4. Comparison of LMCI and EMCI classification performance in the ADNI cohort between the proposed MSRNet and other methods.

Method	ACC (%)	SEN (%)	SPE (%)	AUC (%)
AGNN	67.2	84.5	42.1	75.2
GAT	71.0	81.6	55.1	76.7
GraphSAGE	68.9	75.1	59.8	73.1
HGNN	65.9	85.7	36.9	73.0
Graph U-Net	58.7	57.3	60.4	70.3
ConvMixer	70.7	81.2	55.4	79.9
3D VGG-16	77.2	82.6	69.2	85.7
3DAN	77.3	75.9	79.4	86.0
Proposed MSRNet	79.8	80.4	78.8	87.1

LMCI, late mild cognitive impairment; EMCI, early mild cognitive impairment; ACC, accuracy; SEN, sensitivity; SPE, specificity; AUC, area under the curve; MSRNet, multispatial information representation model; ADNI, Alzheimer’s Disease Neuroimaging Initiative database; AGNN, attention-based graph neural network; GAT, graph attention network; HGNN, hypergraph neural network; 3D VGG-16, 3D visual geometry group 16; 3DAN, 3D attention network.