Machine Learning-Based Model to Predict Heart Disease in Early Stage Employing Different Feature Selection Techniques

Table 12

Compare our predictive results with the previous results.

Authors	Methods	Acc.(%)	Sens. (%)	Spec. (%)	AUROC (%)	Log loss
Our study	SVM and LR	94.51	94.87	94.23	96.08	0.27
Mohan et al. [21]	HRFLM	88.47	92.8	82.6	-	-
Amin et al. [22]	Naïve Bayes and Logistic Regression	87.41	-	-	-	-
Latha & Jeeva [24]	NB, BN, RF, and MP	85.48	-	-	-	-
Patel et al. [23]	J48 with ReducedErrorpruning Algorithm	56.76	-	-	-	-
Tomar & Agarwal [25]	Feature selection-based LSTSVM	85.59	0.8571	0.8913	-	-
Buscema et al. [26]	TWIST algorithm	84.14	-	-	-	-
Subbulakshmi et al. [27]	ELM	87.5	-	-	-	-
Srinivas et al. [28]	Na¨ıve Bayes	83.70	-	-	-	-
Polat & Gunes [29]	Combining of RBF kernel F-score feature selection and LS-SVM classifier	83.70	83.92	83.54	0.831	-
Kahramanli & Allahverdi [30]	Hybrid neural network method	86.8	-	-	-	-