Analysis of arrhythmia detection and classification using electrocardiogram signals with decision tree algorithm

Abstract

Heart disease remains the primary cause of death globally, with arrhythmia diagnosis often limited by restricted access to medical personnel and the complexity of electrocardiogram (ECG) interpretation. Accurate arrhythmia classification is essential to prevent cardiovascular complications. The proposed method successfully categorized classify ECG signals into five categories: normal, abnormal, potentially arrhythmia, moderate arrhythmia risk, and highly potentially arrhythmia. Data were collected from 30 subjects under three activity scenarios: sitting, walking, and running. The proposed model achieved an accuracy of 99.4%, demonstrating strong potential for real-time monitoring applications. Performance evaluation was conducted using accuracy, precision, recall, and F1-score for each class. Although the dataset size remains relatively small, the findings highlight the effectiveness of decision tree as an efficient and interpretable classification method. Future research will involve validation using large-scale public databases like the arrhythmia database at MIT-BIH and comparisons with advanced methods including convolutional neural network (CNN), transformer-based models, and explainable artificial intelligent (XAI) frameworks.