Time Series Analysis using Embedding Dimension on Heart Rate Variability

Abstract Heart Rate Variability (HRV) is the measurement sequence with one or more visible variables of an underlying dynamic system, whose state changes with time. In practice, it is difficult to know what variables determine the actual dynamic system. In this research, Embedding Dimension (ED) is used to find out the nature of the underlying dynamical system. False Nearest Neighbour (FNN) method of estimating ED has been adapted for analysing and predicting variables responsible for HRV time series. It shows that the ED can provide the evidence of dynamic variables which contribute to the HRV time series. Also, the embedding of the HRV time series into a four-dimensional space produced the smallest number of FNN. This result strongly suggests that the Autonomic Nervous System that drives the heart is a two features dynamic system: sympathetic and parasympathetic nervous system.