A New Hybrid Method for Improving the Performance of Myocardial Infarction Prediction

Abstract Introduction: Myocardial Infarction, also known as heart attack, normally occurs due to such causes as smoking, family history, diabetes, and so on. It is recognized as one of the leading causes of death in the world. Therefore, the present study aimed to evaluate the performance of classification models in order to predict Myocardial Infarction, using a feature selection method that includes Forward Selection and Genetic Algorithm. Materials & Methods: The Myocardial Infarction data set used in this study contains the information related to ۵۱۹ visitors to Shahid Madani Specialized Hospital of Khorramabad, Iran. This data set includes ۳۳ features. The proposed method includes a hybrid feature selection method in order to enhance the performance of classification algorithms. The first step of this method selects the features using Forward Selection. At the second step, the selected features were given to a genetic algorithm, in order to select the best features. Classification algorithms entail Ada Boost, Naïve Bayes, J۴۸ decision tree and simpleCART are applied to the data set with selected features, for predicting Myocardial Infarction. Results: The best results have been achieved after applying the proposed feature selection method, which were obtained via simpleCART and J۴۸ algorithms with the accuracies of ۹۶.۵۳% and ۹۶.۳۴%, respectively. Conclusion: Based on the results, the performances of classification algorithms are improved. So, applying the proposed feature selection method, along with classification algorithms seem to be considered as a confident method with respect to predicting the Myocardial Infarction.

Keywords:

Ada Boost , Forward Selection , J۴۸ decision tree , Myocardial Infarction , SimpleCART , کشف دانش , بهینه سازی متوالی کمینه , شبکه عصبی مصنوعی , جنگل تصادفی , سکته قلبی