Analysis and simulation of advanced machine learning methods and deep neural networks for PSVT arrhythmia detection and diagnosis: A review of new feature extraction and classification techniques

Abstract

Paroxysmal supraventricular tachycardia (PSVT) is a common and potentially life-threatening arrhythmia in the cardiac system that begins and ends abruptly. Accurate and timely detection of this arrhythmia is essential to prevent serious complications. In recent years, machine learning methods and deep neural networks have been applied as advanced tools for analyzing electrocardiogram (ECG) signals, which enable non-invasive and automatic identification of PSVT. This paper reviews feature extraction techniques and classification algorithms in PSVT diagnosis. First, feature extraction methods including temporal, frequency, and nonlinear analyses are reviewed. Temporal features include the analysis of statistical indices such as mean and standard deviation, while frequency features are based on Fourier and wavelet transforms. Also, nonlinear analyses such as DFA and RQA help in identifying complex patterns in ECG signals. In the classification algorithms section, traditional models such as Support Vector Machine (SVM), Nearest Neighbor (KNN), and Random Forest (RF) are reviewed. In addition, the application of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) along with transfer learning is discussed. Studies show that combining temporal and frequency features with CNN can improve the accuracy of detection.