Table 2.

Comparison of performance of the Gradient Boosting Decision Tree machine learning (XGBoost) and logistic regression models.

Variable	XGBoosta model			Logistic regression model
	Training, mean (SE)	Test, mean	Training, mean (SE)		Test, mean
Positive predictive value	0.505 (0.099)	0.362	0.441 (0.110)		0.285
AUCb	0.956 (0.015)	0.898	0.943 (0.022)		0.892
Accuracy	0.917 (0.032)	0.918	0.884 (0.049)		0.883
Sensitivity	0.845 (0.021)	0.877	0.874 (0.039)		0.901
Specificity	0.960 (0.016)	0.919	0.946 (0.025)		0.882
F-measure	0.370 (0.107)	0.513	0.306 (0.110)		0.434

^aXGBoost: Gradient Boosting Decision Tree machine learning.

^bAUC: area under the receiver operating characteristic curve.