A Hybrid Intelligent System Framework for the Prediction of Heart Disease Using Machine Learning Algorithms

Table 7

10-fold CV Classification performance of different classifiers on selected features by Relief FS algorithm when .

Predictive model	Classifiers performance evaluation metrics
	Turning parameters	Accuracy (%)	Specificity (%)	Sensitivity (%)	MCC	AUC (%)	Processing time (s)
Logistic regression	C = 1	88	98	76	88	87	16.213
	C = 10	87	98	76	88	87	16.200
	C = 100	89	98	77	89	88	16.111
	C = 0.001	74	98	47	72	73	16.233
K-nearest neighbor	K = 1	80	73	78	80	80	24.400
	K = 3	75	80	72	76	76	24.500
	K = 7	74	78	71	75	75	24.600
	K = 9	73	78	70	75	73	24.611
	K = 13	70	69	71	70	71	21.777
Artificial neural network	16	77	2	100	50	69	21.600
	20	54	96	5	50	68	22.101
SVM (kernel = RBF)	C = 100, = 0.0001	87	95	78	86	87	14.134
	C = 1, = 0.01	79	82	81	79	80	14.139
	C = 10, = 0.001	75	84	68	76	77	14.255
SVM (kernel = linear)	C = 10, = 0.0001	78	95	55	78	74	18.139
	C = 100, = 0.0001	80	97	60	79	79	18.222
Naive Bayes	—	85	87	78	80	84	34.101
Decision tree	100	74	85	66	75	76	20.911
	500	73	84	65	74	74	20.899
Random forest	100	83	93	70	82	83	15.121
	50	85	94	74	82	84	14.330
	25	82	94	70	82	82	14.199