. 2023 Mar 9;20:101116. doi: 10.1016/j.artd.2023.101116

Table 3.

Metrics describing the performance of the best-performing machine learning and control algorithms of the included studies.

Stud, year	Procedure	Outcome	Training size	Validation size	Testing size	Best control algorithm	Control metric, value	Best ML algorithm	ML metric, value
Navarro et al., 2018	TKA	LOS	106,085	-	35,361	-	-	Bayesian	AUC, 0.78
Ramkumar et al., 2019	TKA	LOS	150,074	16,675	4017	-	-	ANN	AUC, 0.83
Ramkumar et al., 2019	THA	LOS	68,810	7645	2771	-	-	ANN	AUC, 0.80
Ramkumar et al., 2019	THA	LOS	91,751	-	30,583	-	-	Bayesian	AUC, 0.87
Lee et al., 2019	Both	90-d readmission	473	52	-	Logistic regression	Accuracy, 0.93	RUSBoost	Accuracy, 0.87
Gabriel et al., 2019	THA	LOS	644	316	-	Logistic regression	AUC, 0.75	Ridge regression	AUC, 0.76
Wei et al., 2021	TKA	LOS	15,069	-	10,046	Logistic regression	AUC, 0.80	ANN	AUC, 0.80
Han et al., 2021	TKA	LOS	1038	260	-	Logistic regression	AUC, 0.70	RFC	AUC, 0.77
Kugelman et al., 2021	THA	LOS	902	225	282	-	-	XGBoost	AUC, 0.82
Yeo et al., 2022	TKA	DOS	6413	1603	2005	-	-	ANN	AUC, 0.82
Klemt et al., 2022	rTKA	LOS	1656	414	518	-	-	ANN	AUC, 0.87
Lopez et al., 2022	TKA	LOS	216,960	-	54,420	-	-	ANN	AUC, 0.80
Lopez et al., 2022	THA	LOS	122,442	-	30,611	-	-	ANN	AUC, 0.81
Abbas et al., 2022	TKA	DOS	182,000	57,841	62,459	Linear regression	MSE, 0.99	ANN	MSE, 0.89
Abbas et al., 2022	TKA	LOS	182,000	57,841	62,459	Linear regression	MSE, 0.79	ANN	MSE, 0.69
Motesharei et al., 2022	TKA	DOS	708	177	176	Linear regression	R2, 0.71	CatBoost	R2, 0.76
Zalikha et al., 2022	TKA	LOS	195,556	48,906	61,115	-	-	SVM	AUC, 0.68
Johannesdottir et al., 2022	Both	LOS	8561	951	-	Logistic regression	AUC, 0.70	RFC	AUC, 0.71
Klemt et al., 2022	TKA	90-day readmission	6413	1603	2005	Logistic regression	-	ANN	AUC, 0.85
Li et al., 2022	TKA	LOS	-	-	-	Logistic regression	AUC, 0.64	XGBoost	AUC, 0.74
Kugelman et al., 2022	TKA	LOS	575	144	180	-	-	XGBoost	AUC, 0.69
Trunfio et al., 2022	THA	LOS	2012	-	503	Linear regression	RMSE, 3.84	GBDT	RMSE, 3.84

GBDT, gradient-boosted decision tree; rTKA, revision total hip arthroplasty; R2, coefficient of determination; RFC, random forest classifier; RMSE, root mean square error; SVM, support vector machines.