Development and comparison of machine learning-based models for predicting heart failure after acute myocardial infarction

Table 2 Diagnostic efficacy of Seven classifiers

Classifier	AUC [95%CI]	Sensitivity	Specifity	Accuracy	Precision	Recall	F1 score	Positive predictive value	Negative predictive value
Decision Tree	0.939 [0.9274–0.9505]	0.875	0.881	0.878	0.877	0.875	0.876	0.877	0.879
AdaBoost	0.940 [0.9289–0.9503]	0.829	0.885	0.857	0.875	0.829	0.851	0.875	0.842
Linear SVC	0.915 [0.9382–0.9584]	0.814	0.869	0.842	0.858	0.814	0.835	0.858	0.828
XgBoost	0.973 [0.9658–0.9797]	0.896	0.955	0.926	0.951	0.896	0.923	0.951	0.905
Random Forest	0.955 [0.9456–0.9646]	0.820	0.952	0.887	0.943	0.820	0.877	0.943	0.845
Random gradient descent	0.906 [0.8916–0.9204]	0.812	0.850	0.831	0.840	0.812	0.825	0.840	0.823
Logistic Regression	0.914 [0.9007–0.9281]	0.810	0.865	0.838	0.854	0.810	0.832	0.854	0.824

*Sensitiity = True Positive /( True Positive + False Negative); Specificity = True Negative/( True Negative + False Positive); Accuracy = (True Positive + True Negative)/( Positive + Negative); Precision = True Positive/( True Positive + False Positive); Recall = True Positive /( True Positive + False Negative); F1 score = = 2*Precision*Recal /(Precision + Recal); Positive predictive value = True Positive/( True Positive + False Positive); Negative predictive value = True Negative/( True Negative + False Negative)

ISSN: 1472-6947