A Discovery Strategy for Active Compounds of Chinese Medicine Based on the Prediction Model of Compound-Disease Relationship

Table 1

Calculation results of the test set based on the GoogLeNet model and the AlexNet model.

CNN	Acc	Sen	Pre	F-measure	MCC	AUC
GoogLeNet	0.960	0.956	0.965	0.960	0.920	0.964
AlexNet	0.919	0.933	0.903	0.918	0.839	0.889

In the same test set, GoogLeNet shows an AUC of 0.964; the AlexNet model shows an AUC of 0.889. Furthermore, the GoogLeNet model has higher values in terms of Acc, Sen, Pre, F-measure, and MCC. It outperforms the AlexNet model by 4.461%, 2.465%, 6.866%, 4.575%, 9.654%, and 8.436%, respectively.