Table 3.

AUROCs of machine learning models trained using data without missing values.

	All variables		Preoperative variables
Machine learning algorithm	AUROC (95% CI) in the test set	AUROC (95% CI) in the validation cohort	AUROC (95% CI) in the test set	AUROC (95% CI) in the validation cohort
Logistic regression	0.844 (0.829–0.859)	0.818 (0.794–0.842)	0.808 (0.791–0.825)	0.811 (0.786–0.835)
Random forest	0.837 (0.821–0.853)	0.813 (0.789–0.837)	0.822 (0.806–0.838)	0.803 (0.778–0.828)
XGBoost	0.839 (0.823–0.855)	0.824 (0.801–0.847)	0.817 (0.801–0.833)	0.792 (0.766–0.818)
GBDT	0.853 (0.838–0.867)	0.830 (0.807–0.852)	0.828 (0.813–0.843)	0.813 (0.789–0.837)

AUROC, area under the receiver operating characteristic curve; GBDT, gradient boosting decision tree; XGBoost, extreme gradient boosting.