Table 3.

Studies of ML-based discharge destination prediction; the “*” denotes the best-performing model

Study	Predicted parameter	Patient population	Methodology	Main factors	Dataset size
Elbattah and Molloy [99]	Discharge destination	Elderly patients with hip fracture care	RF*, Boosted Decision Tree (BDT), NN, Linear regression		(n=2,000)
Karhade et al. [100]	Facility-based discharges	Elective lumbar degenerative disc disorders patients	NN*, BDT, SVM, Bayes Point Machine	Age, sex, BMI, fusion level, functional status	(n= 26,364)
Bacchi et al. [107]	Community-based discharges	Stroke patients	LR*, RF, DT, ANN	Age, sex, estimated pre-stroke mRS	(n= 2,840)
Lu et al. [125]	Facility-based discharges	Unicompartmental knee arthroplasty patients	Generalized linear model, RF, NN, XGB*	Total LOS, preoperative hematocrit, BMI, preoperative sodium	(n=7,275)
Bertsimas et al. [102]	Discharge destination	General patients	LR, CART DT*, Optimal Trees with Parallel Splits, RF, GBDT	Demographics, provider orders, diagnosis codes, medications	(n= 63,432)
Imura et al. [105]	Community-based discharges	Stroke patient	DT, Linear discriminant analysis, KNN*, SVM*, RF	Age, sex, stroke type	(n=481)
Satyadev et al. [103]	Discharge disposition	Traumatic brain injury patients	KNN, XGB, RF*	Vitals, demographics, mechanism of injury, comorbidities	(n=5,292)
Mohammed et al. [104]	Community-based discharges	Total knee arthroplasty patients	LR, GB*, RF*, ANN	Age, sex, race, admission month, admission on a weekend, admission type	(n=572,811)
Bacchi et al. [107]	Community-based discharges	Ischaemic or haemorrhagic stroke patients	LR, ANN*	Age, sex, stroke severity, health history	(n=1,158)
Ikezawa et al. [108]	Community-based discharges	Ischemic cerebral infarction patients	XGB	early nutritional initiation	(n=41,477)
Zhao et al. [133]	Facility-based discharges	Elective radical cystectomy patients	GBDT	Age, race, LOS, BMI	(n=11,881)
Morris et al. [109]	Discharge destination	Elderly patients with trauma	Bayesian additive regression trees	Age, comorbidities, LOS, physiologic parameters	(n=47,037)
Mickle and Deb [110]	Discharge destination	Acute neurological patients	LR, SVM, KNN, XGB*, RF	Age, glucose, admission weight	(n=5,245)
Chen et al. [111]	Facility-based discharges	Total knee arthroplasty patients	ANN*, RF, HGB*, KNN	LOS, age, BMI, sex	(n=434,550)
Geng et al. [137]	Facility-based discharges	Elective anterior cervical discectomy and fusion patients	RF	Age, Medicare insurance, American Society of Anesthesiology score, fusion levels	(n=2,227)