Dynamical Analysis and Data-Driven Modeling of Cardiac Autonomic Activity

Abstract

The heart is a complex organ with several chambers, valves, and tissues that work together to pump blood around the body. Heart rate is the number of beats per minute, and heart rate variability (HRV) measures the fluctuations in the time interval between heartbeats. These fluctuations reflect the activity of the sympathetic and parasympathetic branches of the autonomic nervous system (ANS), which regulate heart rate. Data-driven modeling techniques have been used to extract information from HRV signals, such as HRV time, frequency, and nonlinear domain analysis, and to classify HRV patterns. These methods have been applied to detect cardiovascular diseases, assess stress, and predict mortality and the risk of sudden cardiac death. The HRV of healthy subjects, from neonates to old, is calculated. The neonates showed higher sympathetic activity contribution, which decreased with age until 12, whereas the parasympathetic activity contribution is more. Preadolescence shows higher HRV fluctuations as compared to neonates and older subjects. The HRV of normal and arrhythmia classes, including atrial fibrillation, sudden cardiac death, congestive heart failure, hypertension, and cerebrovascular disease, are computed. In the tune machine learning models, the accuracy of the classifier as NB (71.62%), LR (90.57%), KNN (98.51%), DT (97.07%), RF (99.65%), XG-Boost (99.35) and ANN (99.70%) were achieved. Machine learning can enhance our understanding of HRV and its relation to health and develop more effective methods for assessing and predicting health outcomes. However, further research is needed to overcome the challenges and limitations of applying machine learning to HRV data.