Optimization of Tree-Based Machine Learning Models to Predict the Length of Hospital Stay Using Genetic Algorithm

Table 1

Details of reviewed studies.

Ref.	Sample	LOS type		Predictive method∗			HPO	Best model	Validation indicator
		Continuous	Categorical	Regression	Machine learning	Deep learning
[13]	All admitted patients		—	MR	—	—	—	MR	R ^2	0.267
[14]	ICU-following cardiac surgery patients	—		LR	—	—	—	LR	Accuracy	0.79
[10]	Diabetic patients	—		GLM-NB	RF-GBM	ANN	Various experiments	Ensemble	AUC	0.81
[17]	Newborns	—		NB- LR- MLP	DT-RF-R and.T-SVM	—	Not mentioned	RF	AUC	0.87
[4]	Cardiac patients	—		—	RF-SVM-BN	ANN	Not mentioned	RF	Accuracy	0.8
[12]	Neurosurgery patients		—	—	—	RNN-GRU	—	RNN-GRU	MAE	2.8
[15]	Radical cystectomy for muscle-invasive bladder cancer		—	GLM	—	—	—	GLM	R ^2	0.048
[11]	Brain tumor surgery patients	—		Regression-based models	Tree-based models	ANN	Grid search	Ensemble	RMSE	0.5
[16]	Elective patients	—		NB-BN -KNN- Kstar- LWL	C4.5 DT-SVM-decision table		Not mentioned	BN	AUC	0.9
[18]	ICU	—		FL -KNN-Rerg.T-NB	CT-TB-RF-SVM	ANN	Not mentioned	FL	Accuracy	0.92
[20]	COVID-19 patients		—	—	DT	—	Not mentioned	DT	MAE	2.84
[21]	Inpatients	—		NB- KNN	DT -SVM,-XGBoost- nonlinear weighted XGBoost	—	Grid search	Nonlinear weighted XGBoost	RMSE	1.52
[13]	Lung cancer patients	—		LR	RF- XGBoost	—	Not performed	RF	—	—

Methods. BL, binary logistic; BN, Bayesian network; FL, fuzzy logic; GLM, generalized linear model; KNN, K-nearest neighbors; LR, logistic regression; LWL, locally weighted learning; MLP, multilayer perceptron; MR, multiple regression; NB, Naïve Bayes; Reg.T, regression tree; CT, classification trees; DT, decision tree (J48); GBM, gradient boosting machine; Rand.T, random tree; RF, random forest; SVM, support vector machines; TB, tree bagger; ANN, artificial neural network; GRU, gated recurrent unit; RNN, recurrent neuronal network.