Classification of cardiac arrhythmia using machine learning techniques

Abstract

Cardiovascular disease is one of the leading causes of death worldwide. Currently, there is an increase in the percentage of people with various heart rhythm disorders. There are permanent (chronic), persistent and paroxysmal form of atrial fibrillation, and the most severe violation is the permanent form. Since the reasons for the development of a certain form of atrial fibrillation are not completely clear, this article presents an analysis of various characteristics that affect the formation of arrhythmias of these species. The most significant signs that can potentially be predictors of different forms of the disease have been identified. Four machine learning methods were used for the analysis: Classification trees, logistic regression, random forest, and gradient boosting. The highest cross-validation accuracy was obtained using logistic regression.