Arrhythmia Detection Using ECG-Based Classification with Prioritized Feature Subset Vector-Associated Generative Adversarial Network

Arrhythmia categorization is an exciting research in the early prevention and detection of cardiovascular illnesses, using Electrocardiogram (ECG). In the case of ECG signals, time series data are obtained by changing the time. This type of signal has the drawback of requiring repeated acquisition of comparison data with the same size as the registration data. Resolving the issue of inconsistent data size is accomplished by the use of an additional classifier-based adversarial neural networks. Adversarial data synthesis using Generative Adversarial Networks (GANs) and the generation of additional training examples solves the basic problem of insufficient data labelling. Recent studies have used the GAN architecture to create synthetic adversarial ECG signals in order to boost the amount of training data already available. The arrhythmia detection system, on the other hand, has a fragmented Convolution Neural Network (CNN) classification architecture. No flexible structural design has yet been suggested that can simultaneously discover and order abnormalities. An exceptional Prioritized Feature Subset Vector-Associated Generative Adversarial Network (PFSV-AGAN) is proposed in this research in order at a time produce ECG indications for multiple classes and sense heart-related problems. Furthermore, the model is based on class-specific ECG signals in order to generate realistic adversarial cases. This research presents a framework for ECG signal abnormality identification that has an unbalanced distribution among classes and achieves high accuracy in abnormalities categorization. After training on datasets, the classification model reliably identifies abnormalities in the proposed model. The proposed model when compared to the traditional model exhibits better performance levels. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.