Supraventricular Tachycardia Detection and Classification Model of ECG signal Using Machine Learning

TL;DR

This paper presents a machine learning-based model for detecting and classifying supraventricular tachycardia from ECG signals, achieving 97% accuracy by combining noise filtering, feature extraction, and decision-tree classification.

Contribution

The study introduces a novel ECG feature extraction and classification pipeline specifically for supraventricular tachycardia, highlighting decision trees as the most effective model.

Findings

Decision-tree models outperform other classifiers.

Achieved 97% classification accuracy.

Effective noise filtering improves detection performance.

Abstract

Investigation on the electrocardiogram (ECG) signals is an essential way to diagnose heart disease since the ECG process is noninvasive and easy to use. This work presents a supraventricular arrhythmia prediction model consisting of a few stages, including filtering of noise, a unique collection of ECG characteristics, and automated learning classifying model to classify distinct types, depending on their severity. We de-trend and de-noise a signal to reduce noise to better determine functionality before extractions are performed. After that, we present one R-peak detection method and Q-S detection method as a part of necessary feature extraction. Next parameters are computed that correspond to these features. Using these characteristics, we have developed a classification model based on machine learning that can successfully categorize different types of supraventricular tachycardia. Our findings suggest that decision-tree-based models are the most efficient machine learning models for supraventricular tachycardia arrhythmia. Among all the machine learning models, this model most efficiently lowers the crucial signal misclassification of supraventricular tachycardia. Experimental results indicate satisfactory improvements and demonstrate a superior efficiency of the proposed approach with 97% accuracy.