Effective Computational Modeling for Early Arrhythmia Symptom Classification by Using Decision Tree Approach

Abstract

Heart disease has been the leading global cause of death for almost 15 years. One of the common causes lead to chronic heart disease and sudden death is Arrhythmia. However, the conventional or computational approach of Arrhythmia detection is not an easy task. It requires suitable method with a very specific timeline to detect the symptom. In addition, the symptom itself is very complex in behavior. Therefore, an automatic detection method with simple computational model to detect accurately Arrhythmia in ECG data is needed to deal with such critical issue. In this paper, a novel framework based on decision tree approach by utilizing five peaks taken from ECG segment is proposed to detect Arrhythmia from the first minute of the ECG data. The experimental results show that the proposed decision tree approach with the proposed five peaks is able to detect Arrhythmia with the accuracy of 98% outperforming the other data mining techniques. Moreover, the five proposed parameters to classify the disease show that these computational models have a strong level of sustainability in detecting Arrhythmia when it is compared to different numbers of parameters and methods.