Table 4:

Comparison of model performance on WRAT classification task

	WRAT classification with nback fMRI
	Method	ACC	SEN	SPE	AUC
Graph based models	TAG	0.6084 ± 0.0253	0.6494 ± 0.0363	0.5713 ± 0.0360	0.6574 ± 0.0276
	SAGE	0.6427 ± 0.0258	0.4564 ± 0.0395	0.8134 ± 0.0295	0.7099 ± 0.0274
	GCN	0.6432 ± 0.0262	0.5202 ± 0.0361	0.7691 ± 0.0334	0.7094 ± 0.0279
	CHEB	0.6448 ± 0.0249	0.7623 ± 0.0305	0.5131 ± 0.0373	0.6876 ± 0.0283
	GAT	0.6967 ± 0.0246	0.7147 ± 0.0352	0.6815 ± 0.0340	0.7149 ± 0.0276
Hypergraph based models	HGSVM-L	0.6430 ± 0.0253	0.5833 ± 0.0381	0.6972 ± 0.0333	0.7168 ± 0.0271
	HGSVM-R	0.6511 ± 0.0245	0.5978 ± 0.0371	0.6995 ± 0.0332	0.6960 ± 0.0273
	HGNN	0.6531 ± 0.0238	0.6975 ± 0.0330	0.6051 ± 0.0369	0.7150 ± 0.0268
	wHGNN	0.6851 ± 0.0241	0.7314 ± 0.0317	0.6301 ± 0.0379	0.7648 ± 0.0238
	dw HGCN (ours)	0.7149 ± 0.0245	0.7597 ± 0.0314	0.6614 ± 0.0375	0.7760 ± 0.0242
	WRAT classification with emoid fMRI
	Method	ACC	SEN	SPE	AUC
Graph based models	GCN	0.6203 ± 0.0256	0.5591 ± 0.0373	0.6846 ± 0.0345	0.6606 ± 0.0281
	SAGE	0.6276 ± 0.0266	0.6232 ± 0.0396	0.6315 ± 0.0345	0.6410 ± 0.0302
	CHEB	0.6371 ± 0.0268	0.6235 ± 0.0375	0.6512 ± 0.0367	0.6828 ± 0.0293
	TAG	0.6428 ± 0.0254	0.6619 ± 0.0356	0.6250 ± 0.0352	0.6885 ± 0.0279
	GAT	0.6606 ± 0.0268	0.6185 ± 0.0390	0.6956 ± 0.0345	0.6892 ± 0.0291
Hypergraph based models	HGSVM-R	0.6105 ± 0.0252	0.5378 ± 0.0367	0.6772 ± 0.0356	0.6508 ± 0.0287
	HGNN	0.6123 ± 0.0260	0.6701 ± 0.0354	0.5559 ± 0.0374	0.6544 ± 0.0291
	HGSVM-L	0.6191 ± 0.0258	0.5882 ± 0.0392	0.6469 ± 0.0362	0.6369 ± 0.0287
	wHGNN	0.6208 ± 0.0259	0.6612 ± 0.0357	0.5768 ± 0.0378	0.6534 ± 0.0292
	dw HGCN (ours)	0.6485 ± 0.0267	0.7852 ± 0.0333	0.5158 ± 0.0374	0.7173 ± 0.0279