Table 2.

The performance comparison between NHGNN-DTA and other SOTA models on the KIBA dataset.

Method	MSE $\downarrow$	CI $\uparrow$	$r_m^2\uparrow$
DeepDTA	0.194(0.008)	0.863(0.005)	0.673(0.019)
MT-DTI	0.152	0.882	0.738
GraphDTA	0.139(0.008)	0.891(0.001)	0.725(0.018)
rzMLP	0.142	0.89	0.748
EnsembleDLM	0.138(0.003)	0.895(0.001)
FusionDTA	0.130(0.002)	0.906(0.001)	0.793(0.002)
MgraphDTA	0.128(0.001)	0.902(0.001)	0.801(0.001)
NHGNN(Ours)	0.124(0.002)	0.907(0.001)	0.807(0.002)

Bold corresponds to the best performance for each metric, and underline indicates the second best.