Table 4.

Clinical outcome of interests with evaluation metric(s) and prediction period level.

Evaluation Tasks (# paper)	Evaluation Metrics	Time periods	Papers
Disease Prediction (31)	AUROC, AUPRC, Accuracy, Top K Recall/Precision	Year-level, visit-level	[6,11,27,31,33,38,43,54,56,57,59,60,62–64,66,67,69–72,75,76,80,81,83,86,89–91,98]
Mortality Prediction (12)	AUROC, AUPRC, F1, Accuracy	24-hour, month-level	[6,14,43,57,58,61,65,73,75,77,85,98]
Length-of-stay Prediction (8)	AUROC, AUPRC, F1, Accuracy, MSE, MAPE	12-hour, 24-hour	[6,14,43,73,82,83,85,87]
Admission Prediction (7)	AUROC, AUPRC, Accuracy, Top K Recall/Precision	Month-level, year-level	[6,38,48,77–79,88]
Patient Subtyping (4)	Significant difference among different groups; NMI, MSE, Sparsity and Similarity	Visit-level	[67,70,74,79]
Intervention Prediction (5)	Jaccard, Recall, Precision, F1	24-hour	[32,54,55,65,98]
Medical Cost (3)	R-squared, RMSE	Year-level	[48,85,87]