TABLE 3.

Classification performance of each model evaluated externally on the community dataset.

Method	Sensitivity	Specitivity	Accuracy	AUROC	Precision	F1-score	AUPRC
Logistic regression	0.6215	0.8895	0.8464	0.8435	0.5189	0.5656	0.5199
SVM_l	0.5763	0.9458	0.8864	0.9282	0.6711	0.6201	0.6652
SVM_r	0.6102	0.9404	0.8873	0.9137	0.6626	0.6353	0.6395
SVM_s	0.5650	0.9437	0.8827	0.9177	0.6579	0.6079	0.6560
SVM_p	0.6045	0.9415	0.8873	0.9213	0.6646	0.6331	0.6549
Neural network	0.5876	0.9426	0.8855	0.9139	0.6624	0.6228	0.6513
Random forest	0.5706	0.9437	0.8836	0.9259	0.6601	0.6121	0.6623
XGBoost	0.5424	0.9415	0.8773	0.9006	0.6400	0.5872	0.6323
LASSO	0.5932	0.9393	0.8836	0.9023	0.6522	0.6213	0.6284
Best subset	0.4859	0.9274	0.8564	0.8483	0.5621	0.5212	0.5432