Table 1.

Comparison of model performance between logistic regression and machine-learning models.

Duration	Logistic regression		Gradient boosting machine		Deep learning		Random forest
	AUCa (range)	RMSEb	AUC (range)	RMSE	AUC (range)	RMSE	AUC (range)	RMSE
3 years	0.7401 (0.7262-0.7541)	0.1203	0.7927 (0.7803-0.8051)	0.1197	0.7769 (0.7639-0.7899)	0.1244	0.7868 (0.7742-0.7993)	0.1198
5 years	0.7192 (0.7084-0.7301)	0.1633	0.7769 (0.7673-0.7864)	0.1620	0.7610 (0.7566-0.7762)	0.1667	0.7769 (0.7612-0.7804)	0.1622
7 years	0.6990 (0.6901-0.7077)	0.2087	0.7589 (0.751-0.7668)	0.2063	0.7526 (0.7446-0.7606)	0.2099	0.7531 (0.7452-0.761)	0.2066
10 years	0.6885 (0.6801-0.6961)	0.2318	0.7491 (0.7426-0.7570 )	0.2314	0.7374 (0.7339-0.7486)	0.2435	0.7439 (0.7365-0.7510)	0.2318

^aAUC: area under the receiver operating characteristic curve.

^bRMSE: root mean squared error.