Table 5.

Studies related to other ML-based discharge outcomes prediction; the “*” denotes the best-performing model

Study	Predicted parameter	Patient population	Methodology	Main factors	Dataset size
Ebell et al. [140]	Survival to discharge	Cardiopulmonary patients	Classification, Regression trees	Demographics, clinical features at admission	(n = 38,092)
Morton et al. [124]	Prolonged LOS	Diabetic patients	MLR, SVM*, SVM+, MTL, RF	Age, Sex, Race, Expected Primary Payer, Admission Type	(n= 10,000)
Luo et al. [125]	Daily discharges	Nephrology patients	Time series (ARIMA, LSTM and RF*	Demographics, discharge date	(n=1,091)
McCoy Jr et al. [35]	Discharge volume	General patients	Time-series		(n=101,867)
Van Walraven et al. [126]	Daily discharges	General patients	Survival tree approach	Age, sex, patient location throughout the admission	(n= 192,859)
Levin et al. [128]	Discharge rounds	General patients	Unit- specific models	Demographics, administrative, medications	(n= 12,470)
Ghazalbash et al. [129]	Multimorbidity status of patients	Old patients with discharge delay	Classification and regression trees, RF*, Bagging trees, XGB*, LR	Age, sex, marginalization, rural/urban residency, chronic conditions, LOS, admission type	(n=163,983)
Ahn et al. [131]	Discharge probability	Cardiovascular patients	XGB*, LR, RF, SVM, Multilayer perceptron	Demographics, administrative, medications	(n=572,811)
Gramaje et al. [130]	Discharge or Remain	Surgical patients	DT, RF, Bayesian Network*	Age, clinical conditions	(n=90)
Gao et al. [132]	Inpatient discharge	General patients	edRVFL		(n=417)
Jaotombo et al. [134]	LOS	General patients	LR, CART, RF, GB*, NN	Discharge destination, age, emergency admission, with more comorbidities notably mental health problems and dementia	(n=73,182)