Table 2.

Average area under the receiver operating characteristic curve and its standard deviation for all experiments.

	Model	Center A (n = 1,160)		Center B (n = 631)		Average of centers
		Balanced class weight	Random oversampling	Balanced class weight	Random oversampling	Balanced class weight	Random oversampling
Cyclical	XGBoost	0.58 ± 0.10	0.58 ± 0.10	0.62 ± 0.16	0.54 ± 0.15	0.60 ± 0.13	0.56 ± 0.13
	CatBoost	0.62 ± 0.15	0.60 ± 0.14	0.61 ± 0.14	0.61 ± 0.16	0.62 ± 0.15	0.61 ± 0.15
	Random forest	0.62 ± 0.11	0.61 ± 0.12	0.64 ± 0.13	0.64 ± 0.14	0.63 ± 0.12	0.63 ± 0.13
	Neural network (wide)	0.62 ± 0.14	0.63 ± 0.14	0.67 ± 0.14	0.65 ± 0.17	0.65 ± 0.14	0.64 ± 0.16
	Neural network (narrow)	0.64 ± 0.12	0.62 ± 0.13	0.68 ± 0.12	0.62 ± 0.15	0.66 ± 0.12	0.62 ± 0.14
Stacking	XGBoost	0.67 ± 0.10	0.61 ± 0.08	0.63 ± 0.17	0.60 ± 0.13	0.65 ± 0.14	0.61 ± 0.11
	CatBoost	0.64 ± 0.11	0.62 ± 0.10	0.65 ± 0.16	0.62 ± 0.13	0.65 ± 0.14	0.62 ± 0.12
	Random forest	0.63 ± 0.10	0.60 ± 0.09	0.64 ± 0.15	0.63 ± 0.15	0.64 ± 0.13	0.62 ± 0.12
	Neural network (wide)	0.64 ± 0.13	0.62 ± 0.13	0.64 ± 0.14	0.61 ± 0.11	0.64 ± 0.14	0.62 ± 0.12
	Neural network (narrow)	0.64 ± 0.12	0.65 ± 0.13	0.66 ± 0.14	0.59 ± 0.14	0.65 ± 0.13	0.62 ± 0.14
Mono-center	XGBoost	0.65 ± 0.11	0.59 ± 0.11	0.59 ± 0.17	0.56 ± 0.18	0.62 ± 0.14	0.58 ± 0.15
	CatBoost	0.63 ± 0.11	0.59 ± 0.12	0.60 ± 0.15	0.64 ± 0.17	0.62 ± 0.13	0.62 ± 0.15
	Random forest	0.65 ± 0.10	0.59 ± 0.11	0.62 ± 0.14	0.62 ± 0.16	0.64 ± 0.12	0.61 ± 0.14
	Neural network (wide)	0.64 ± 0.11	0.62 ± 0.13	0.63 ± 0.15	0.61 ± 0.15	0.64 ± 0.13	0.62 ± 0.14
	Neural network (narrow)	0.63 ± 0.12	0.58 ± 0.12	0.65 ± 0.16	0.60 ± 0.16	0.64 ± 0.14	0.59 ± 0.14

The rows are the classifiers on different setups (cyclical, stacking or internal validation) and the columns are different balancing techniques per center. Highest accuracies per center and on average are highlighted in bold.