Abstract
Dynamic correlation is the correlation between two time series across time. Two approaches that currently exist in neuroscience literature for dynamic correlation estimation are the sliding window method and dynamic conditional correlation. In this paper, we first show the limitations of these two methods especially in the presence of extreme values. We present an alternate approach for dynamic correlation estimation based on a weighted graph and show using simulations and real data analyses the advantages of the new approach over the existing ones. We also provide some theoretical justifications and present a framework for quantifying uncertainty and testing hypotheses.
Highlights
IntroductionDynamic correlation for short, is correlation between a pair of time series, which itself changes over time
Dynamic bivariate correlation, or dynamic correlation for short, is correlation between a pair of time series, which itself changes over time
As T got larger, the dynamic conditional correlation (DCC) performance overtook that of weighted graph algorithm (WGA), and at T = 1000, DCC performed better than WGA based on the metrics used in Table
Summary
Dynamic correlation for short, is correlation between a pair of time series, which itself changes over time. Assessing dynamic correlation is of importance in many areas within neuroscience. Dynamic correlation estimation is of interest in neuroimaging because many studies have identified dynamic changes in functional connectivity during the course of a functional magnetic resonance imaging (fMRI) experiment [1,2,3,4,5,6,7], especially during resting state. Dynamic correlation between local field potential time series obtained from different brain regions could be used to explore how certain brain regions work in tandem during certain specific behaviors. When identifying such changes, it is of importance to make sure that the dynamic shifts in the correlations observed are not due to spurious fluctuations inherent to the estimation method used
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have