Conventional and Heterogeneous Cascade Reservoir Computing for Arrhythmia Detection: A Comparative Study

Cardiovascular diseases are considered the pri-mary cause of mortality on a global scale according to the World Health Organization (WHO). As a result of the simplicity, safety, and capability of capturing heart conditions, the electrocardio-gram (ECG) signal is extensively employed to tackle the issue of cardiovascular diseases (CVD). In this study, we performed a comparative study between two reservoir computing (RC)-based arrhythmia classifiers. A conventional R C and a cascade one. We cascade a typical RC with a next-generation RC (NG-RC). This study is assessed utilizing the MIT-BIH database according to the guidelines recommended by the Association for the Advancement of Medical Instrumentation (AAMI). Here, we will consider intra- and inter-patient paradigms. The outcomes achieved by implementing the proposed classifiers convention RC and cascade RCs are an overall accuracy of 98.67% and 98.89% for intra-patient scenarios and 95.98% and 95.21 % for inter-patient cases, respectively. These results show that the cascade-based classifier enhances classification accuracy compared to the typical RC-based detector. However, NG-RC-based RCs don't perform well in the classification context when the inter-patient paradigm is considered.