Table 2.

Performance comparisons for different learning algorithms on derivation and temporal sets. The values in the cell of the top table are the mean and 1.96 standard deviations of 10-fold cross-validation. Random forest and XGBoost both work well, and RF outperforms XGBoost in AUC.

Performance on Derivation Set
Method	AUC	Specificity	Sensitivity
RF	0.787 ± 0.185	0.955 ± 0.187	0.653 ± 0.334
XGBoost	0.782 ± 0.268	0.905 ± 0.335	0.729 ± 0.329
Decision Tree	0.576 ± 0.229	0.698 ± 0.349	0.517 ± 0.335
Logistic Regression	0.538 ± 0.273	0.717 ± 0.230	0.695 ± 0.250
Performance on Temporal Set
Method	AUC	Specificity	Sensitivity
RF	0.771	0.815	0.5
XGBoost	0.759	0.796	0.5
Decision Tree	0.632	0.888	0.25
Logistic Regression	0.671	0.870	0.5