Table 2.

The detection performance of 9 algorithms for the internal validation dataset.

Algorithms	Accuracy	Sensitivity	Specificity	PPVa	NPVb	Brier score	AUCc
SVMd	0.849	0.377	0.985	0.880	0.845	0.151	0.766
Random forest	0.833	0.432	0.949	0.709	0.852	0.167	0.728
AdaBoost	0.860	0.376	1	1	0.847	0.140	0.763
kNNe	0.841	0.415	0.964	0.768	0.851	0.159	0.723
NBf	0.845	0.367	0.983	0.860	0.843	0.155	0.768
Decision tree	0.838	0.431	0.956	0.738	0.853	0.162	0.706
LRg	0.844	0.363	0.984	0.865	0.842	0.156	0.765
XGBoosth	0.860	0.377	1	1	0.847	0.140	0.771
GBDTi	0.860	0.376	1	1	0.847	0.140	0.772

^aPPV: positive predictive value.

^bNPV: negative predictive value.

^cAUC: area under the curve.

^dSVM: support vector machine.

^ekNN: k-nearest neighbors.

^fNB: naive Bayes.

^gLR: logistic regression.

^hXGBoost: eXtreme gradient boosting.

ⁱGBDT: gradient boosting decision tree.