Table 3.

Evaluation indicators of predictive performance of five models in HER2-overexpressed and Non-HER2-overexpressed groups.

Classifier		SEN	SPE	PRE	GM	FPR	F1	ACC	AUC
LR	Training Dataset	0.5000	0.7926	0.5232	0.5744	0.2074	0.5067	0.7026	0.7068
	Testing Dataset	0.8462	0.3529	0.7500	0.5465	0.6471	0.7952	0.6964	0.7029
RF	Training Dataset	0.3667	0.9704	0.9449	0.5782	0.0296	0.5105	0.7862	0.8065
	Testing Dataset	0.2941	0.9744	0.8333	0.5353	0.0256	0.4348	0.7679	0.8054
NB	Training Dataset	0.3833	0.8296	0.5958	0.5338	0.1704	0.4308	0.6923	0.6932
	Testing Dataset	0.8718	0.3529	0.7556	0.5547	0.6471	0.8095	0.7143	0.7164
SVM	Training Dataset	0.3667	0.8963	0.8269	0.5580	0.1037	0.4563	0.7333	0.7883
	Testing Dataset	0.8974	0.4118	0.7778	0.6079	0.5882	0.8333	0.7500	0.7617
XGBoost	Training Dataset	0.5167	0.9185	0.7972	0.6341	0.0815	0.6049	0.7949	0.7988
	Testing Dataset	0.7949	0.6471	0.8378	0.7172	0.3529	0.8158	0.7500	0.7459

RF, Random Forest; NB, Naïve Bayes; SVM, Support Vector Machine; XGBoost, eXtreme Gradient Boosting; SEN, sensitivity ;SPE, specificity; PRE, precision; GM, geometric mean; FPR, false positive rate; ACC, accuracy; AUC, area under ROC. LR, Logistic Regression.