Table 2.

Performance evaluation metrics for prediction of patient admission/readmission.

	Sampling type	Accuracy score	Precision Score	Recall score	F1 score	AUC
Patient admission
Decision tree	WS	0.93	0.70	0.75	0.72	0.85
	OS	0.92	0.64	0.83	0.72	0.88
	US	0.91	0.60	0.92	0.72	0.91
Random forest	WS	0.95	0.82	0.79	0.81	0.98
	OS	0.94	0.72	0.87	0.79	0.98
	US	0.93	0.68	0.93	0.78	0.98
Logistic regression	WS	0.91	0.76	0.44	0.56	0.94
	OS	0.92	0.63	0.94	0.75	0.96
	US	0.93	0.70	0.80	0.75	0.96
Support vector machine	WS	0.95	0.80	0.79	0.80	0.97
	OS	0.92	0.62	0.97	0.76	0.98
	US	0.94	0.70	0.94	0.81	0.98
Patient readmission
Decision tree	WS	0.55	0.52	0.62	0.56	0.52
Random forest	WS	0.55	0.52	0.60	0.56	0.58
Logistic regression	WS	0.55	0.52	0.46	0.49	0.48
Support vector machine	WS	0.52	0.49	0.48	0.49	0.50

Distribution type: without sampling (WS), oversampling (OS), undersampling (US). Best performing model in bold.