Table 1.

The performance comparison between NHGNN-DTA and other SOTA models on the Davis dataset.^a

Method	MSE $\downarrow$	CI $\uparrow$	$r_m^2\uparrow$
DeepDTA	0.261(0.007)	0.878(0.002)	0.63(0.015)
MT-DTI	0.245	0.887	0.665
GraphDTA	0.229(0.005)	0.893(0.002)	0.685(0.016)
GEFA	0.228	0.893
rzMLP	0.205	0.896	0.709
EnsembleDLM	0.202(0.005)	0.907(0.004)
FusionDTA	0.208(0.002)	0.913(0.001)	0.743(0.002)
MgraphDTA	0.207(0.001)	0.900(0.004)	0.710(0.005)
NHGNN(Ours)	0.196(0.004)	0.914(0.002)	0.744(0.003)

^aBold corresponds to the best performance for each metric, and underline indicates the second best. $\uparrow$/$\downarrow$ indicates that the larger/smaller the metrics, the better the model performance.