Assessment of Sympathetic and Parasympathetic Activities of Nervous System from Heart Rate Variability Using Machine Learning Techniques

Abstract

The heart is the major organ of the circulatory system, maintaining blood flow and nutrition transfer to various cells and tissues. Heart rate variability (HRV) is the time fluctuation between successive heartbeats. In the past 2 decades, HRV research has gained importance and has produced some of the most promising illness indicators. A well-known non-invasive method to determine how the autonomic nervous system works is the analysis of HRV. The sympathetic and parasympathetic nervous systems are essential for information transfer to the autonomic nervous system (ANS). The clinical development of many diseases is significantly influenced by dysfunction of the autonomic nerve system. The research focuses on the sympathetic nervous systems (SNS) and parasympathetic nervous system’s (PNS) role in HRV regulation. This research is conducted on the HRV data of middle-aged adults. The people were put under different stress levels, and ECG data are collected from them. The data are often imbalanced, so data balancing is provided in this work. The aim of the research is to assess the ANS dysfunction using HRV. Both linear and nonlinear parameters of HRV are used here to assess the ANS. 36 features are extracted from ECG and feature reduction/selection technique is designed to reduce these numbers of features. Machine learning models are developed to assess ANS activity. Our findings from analysis can be stated as “Lower HRV levels suggest the sight of ANS inadequacy, while higher HRV values demonstrate effective ANS”. The performance is assessed using accuracy of the proposed model against other current approaches. The results show that the proposed model achieved 99% accuracy.