Developing an artificial intelligence method for screening hepatotoxic compounds in traditional Chinese medicine and Western medicine combination

Table 1 AI models for drug-included liver injury

Dataset	ML method	AUC	CA	F1	Precision	Recall
Training set	SGD	0.647	0.709	0.715	0.723	0.709
	kNN	0.654	0.722	0.698	0.690	0.722
	SVM	0.785	0.795	0.791	0.788	0.795
	DT	0.710	0.756	0.760	0.764	0.756
	RF	0.814	0.838	0.827	0.832	0.838
	Adaboost	0.785	0.792	0.788	0.785	0.792
	ANN	0.621	0.737	0.671	0.685	0.737
	LR	0.746	0.776	0.757	0.757	0.761
	NB	0.632	0.686	0.694	0.711	0.686
Test set	SGD	0.627	0.682	0.694	0.712	0.682
	kNN	0.574	0.745	0.636	0.555	0.745
	SVM	0.669	0.747	0.712	0.710	0.747
	DT	0.544	0.680	0.679	0.678	0.680
	RF	0.739	0.767	0.731	0.739	0.767
	Adaboost	0.614	0.708	0.707	0.707	0.708
	ANN	0.647	0.694	0.696	0.697	0.694
	LR	0.656	0.733	0.694	0.688	0.733
	NB	0.598	0.675	0.648	0.705	0.675

LR Logistic regression, RF Random forest, SVM Support vector machine, kNN k-nearest neighbor, DT Decision tree, NB Naive bayes, ANN Artificial neural network, SGD Stochastic gradient descent

ISSN: 1749-8546