Application of machine learning missing data imputation techniques in clinical decision making: taking the discharge assessment of patients with spontaneous supratentorial intracerebral hemorrhage as an example

Table 7 The P values of statistical test between EL and other methods

Missing mechanisms	Evaluation metrics	P values	Machine learning methods				Traditional methods
Missing mechanisms	Evaluation metrics	P values	LR	RF	NN	SVM	Mode	KNN	MICE
MCAR	Sensitivity	p.raw	0.018	0.047	0.091	0.016	0.016	0.016	0.016
		p.adj	0.025	0.055	0.091	0.025	0.025	0.025	0.025
	AUC	p.raw	0.016	0.029	0.139	0.018	0.016	0.030	0.018
		p.adj	0.031	0.034	0.139	0.031	0.031	0.034	0.031
	Kappa	p.raw	0.018	0.016	0.050	0.016	0.030	0.016	0.016
		p.adj	0.025	0.025	0.050	0.025	0.034	0.025	0.025
MAR (the ratio of missing proportion 1:2)	Sensitivity	p.raw	0.028	0.050	0.091	0.030	0.016	0.030	0.016
		p.adj	0.041	0.059	0.091	0.041	0.041	0.041	0.041
	AUC	p.raw	0.029	0.017	0.029	0.018	0.030	0.029	0.018
		p.adj	0.030	0.030	0.030	0.030	0.030	0.030	0.030
	Kappa	p.raw	0.024	0.050	0.091	0.029	0.953	0.086	0.016
		p.adj	0.068	0.088	0.106	0.068	0.953	0.106	0.068
MAR (the ratio of missing proportion 2:1)	Sensitivity	p.raw	0.016	0.018	0.172	0.017	0.016	0.018	0.016
		p.adj	0.021	0.021	0.172	0.021	0.021	0.021	0.021
	AUC	p.raw	0.050	0.584	0.819	0.086	0.016	0.071	0.031
		p.adj	0.117	0.681	0.819	0.120	0.109	0.120	0.109
	Kappa	p.raw	0.016	0.016	0.016	0.018	0.016	0.016	0.016
		p.adj	0.018	0.018	0.018	0.018	0.018	0.018	0.018

ISSN: 1472-6947