Feature Selection and Dwarf Mongoose Optimization Enabled Deep Learning for Heart Disease Detection

Table 1

Comparative discussion.

Variations	Metrics	BF-PSO	Bi-LSTM-CRF	XGBoost	RLNNC	DMOA-SqueezeNet(without feature selection)	Proposed DMOA-SqueezeNet
Cleveland dataset
Training data	Accuracy	0.900	0.906	0.911	0.918	0.919	0.925
	Sensitivity	0.835	0.857	0.883	0.908	0.912	0.926
	Specificity	0.879	0.891	0.902	0.917	0.915	0.918
K value	Accuracy	0.907	0.911	0.915	0.918	0.919	0.922
	Sensitivity	0.896	0.905	0.905	0.912	0.915	0.918
	Specificity	0.853	0.865	0.876	0.890	0.898	0.909
Z-Alizadeh Sani dataset
Training data	Accuracy	0.886	0.892	0.897	0.903	0.904	0.911
	Sensitivity	0.827	0.848	0.874	0.899	0.902	0.917
	Specificity	0.870	0.882	0.893	0.907	0.907	0.908
K value	Accuracy	0.887	0.890	0.894	0.897	0.898	0.902
	Sensitivity	0.885	0.894	0.894	0.901	0.903	0.907
	Specificity	0.846	0.859	0.869	0.884	0.891	0.903