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 4 Evaluation results of different processing methods in different scenarios of MCAR mechanism

Evaluation metrics	Missing proportions	Machine learning methods					Traditional methods
Evaluation metrics	Missing proportions	LR	RF	NN	SVM	EL	Mode	KNN	MICE
Sensitivity	0.05	0.874	0.874	0.877	0.874	0.877	0.854	0.874	0.870
	0.10	0.889	0.881	0.881	0.877	0.893	0.847	0.877	0.877
	0.15	0.866	0.866	0.885	0.866	0.889	0.835	0.866	0.862
	0.20	0.877	0.874	0.893	0.866	0.893	0.851	0.866	0.872
	0.30	0.877	0.870	0.885	0.866	0.900	0.839	0.866	0.868
	0.50	0.847	0.904	0.893	0.862	0.893	0.793	0.851	0.849
	Average	0.872	0.878	0.886	0.869	0.891	0.837	0.867	0.866
AUC	0.05	0.912	0.913	0.914	0.913	0.915	0.911	0.913	0.912
	0.10	0.921	0.917	0.918	0.915	0.922	0.908	0.916	0.915
	0.15	0.908	0.914	0.918	0.914	0.915	0.895	0.915	0.907
	0.20	0.908	0.916	0.918	0.913	0.918	0.901	0.913	0.915
	0.30	0.909	0.915	0.916	0.913	0.926	0.893	0.914	0.913
	0.50	0.892	0.923	0.922	0.910	0.923	0.877	0.901	0.894
	Average	0.908	0.916	0.918	0.913	0.920	0.898	0.912	0.909
Kappa	0.05	0.553	0.553	0.555	0.553	0.555	0.555	0.553	0.551
	0.10	0.566	0.561	0.561	0.559	0.568	0.557	0.559	0.558
	0.15	0.552	0.552	0.564	0.552	0.566	0.497	0.553	0.545
	0.20	0.568	0.566	0.578	0.561	0.578	0.532	0.563	0.566
	0.30	0.562	0.557	0.566	0.555	0.576	0.512	0.560	0.574
	0.50	0.533	0.569	0.562	0.543	0.596	0.493	0.540	0.524
	Average	0.556	0.560	0.564	0.554	0.573	0.524	0.555	0.553

ISSN: 1472-6947