Improving an Intelligent Detection System for Coronary Heart Disease Using a Two-Tier Classifier Ensemble

Table 8

Comparison of the proposed method with some prior studies using the Z-Alizadeh Sani dataset (the best value is indicated in bold).

Study	Technique	# of features	Validation method	Accuracy	(%)	(%)	AUC (%)	Statistical test
[16]	Bagging-DT	20	10CV	79.54; 61.46 and (RCA)	(Lad), (LCX), 68.96	Not reported	Not reported	No
[17]	Information gain-SMO	34	10CV	94.08		Not reported	Not reported	No
[18]	Information gain-SVM	24	10CV	86.14; 83.17 and 83.5	(Lad), (LCX), (RCA)	Not reported	Not reported	No
[21]	Neural network genetic algorithm	22	10CV	93.85		Not reported	Not reported	No
[20]	Ensemble algorithm multiple feature selection	34	10CV	93.70		95.53	Not reported	No
[46]	Support vector machine feature engineering	32	10CV	96.40		Not reported	Not reported	No
[25]	-support vector machine	29	10CV	93.08		91.51	Not reported	No
This paper	Two-tier ensemble PSO-based feature selection	27	10CV	98.13		96.60	98.70	Two-step statistical test