Information Decomposition in the Frequency Domain: a New Framework to Study Cardiovascular and Cardiorespiratory Oscillations

Abstract

While cross-spectral and information-theoretic approaches are widely used for the multivariate analysis of physiological time series, their combined utilization is far less developed in the literature. This study introduces a framework for the spectral decomposition of multivariate information measures, which provides frequency-specific quantifications of the information shared between a target and two source time series and of its expansion into amounts related to how the sources contribute to the target dynamics with unique, redundant and synergistic information. The framework is illustrated in simulations of linearly interacting stochastic processes, showing how it allows to retrieve amounts of information shared by the processes within specific frequency bands which are otherwise not detectable by time-domain information measures, as well as coupling features which are not detectable by spectral measures. Then, it is applied to the time series of heart period, systolic and diastolic arterial pressure and respiration variability measured in healthy subjects monitored in the resting supine position and during head-up tilt. We show that the spectral measures of unique, redundant and synergistic information shared by these variability series, integrated within specific frequency bands of physiological interest, reflect the mechanisms of short term regulation of cardiovascular and cardiorespiratory oscillations and their alterations induced by the postural stress.

Multivariate Correlation Measures Reveal Structure and Strength of Brain–Body Physiological Networks at Rest and During Mental Stress

“…Such protocol should also include intermediate resting phases between stressful situations to assess whether elicited stress still produces effects during time in a consequent resting phase. Future methodological work is also envisaged regarding: (a) a thorough validation on simulations of the MI measures presented here performed also through a direct comparison with more sophisticated analysis techniques including the use of time-delayed techniques employing tools of information dynamics to retrieve directional information ( Faes et al, 2014 ) and of non-linear model free entropy estimators ( Faes et al, 2015a ); (b) the frequency-specific decomposition of the proposed measures (e.g., following Faes et al, 2020 ) to investigate how MIs can reflect oscillatory rhythms with specific physiological meaning; and (c) the analysis on source-reconstructed signals to obtain better anatomically-localized estimates of the strength and topology of brain–body interactions ( Lai et al, 2018 ; Van de Steen et al, 2019 ; Kotiuchyi et al, 2020 ).…”

Information theoretic evidence for layer- and frequency-specific changes in cortical information processing under anesthesia

“…Another potential method comes from the equality of Granger causality and information transfer or information theoretic measure for Gaussian variables in the time and frequency domain [73,74]. Also, recent work showed that for linear Gaussian processes the information modification (one of the component of information processing) may be formulated, analytically, in the frequency domain [75,76] as the synergy component of a partial information decomposition following the idea of [77], and using the PID measure I MMI of [78]. This frequency decomposition approach for linear Gaussian processes can be also adopted for the AIS.…”

“…This frequency decomposition approach for linear Gaussian processes can be also adopted for the AIS. When the assumptions of linear Gaussian processes are valid, then the methods in [71] or [75,76], will be more data-efficient and come with lower computational burden. Furthermore, the frequency estimation of the AIS for Gaussian variables will then allow for a more precise identification of the relevant frequency components (the work in [75,76], evaluates the all frequency spectrum while the current method can only identify frequency bands).…”

“…When the assumptions of linear Gaussian processes are valid, then the methods in [71] or [75,76], will be more data-efficient and come with lower computational burden. Furthermore, the frequency estimation of the AIS for Gaussian variables will then allow for a more precise identification of the relevant frequency components (the work in [75,76], evaluates the all frequency spectrum while the current method can only identify frequency bands). If the assumptions of linear Gaussian processes are not met however, our approach seems to be the only viable alternative at the moment.…”

Spectral decomposition of cerebrovascular and cardiovascular interactions in patients prone to postural syncope and healthy controls

“…In this work, we introduce a framework for the evaluation of total, directed and instantaneous interactions between pairs of time series in both time and frequency domains. The framework is built on previous works establishing a connection between information-theoretic and spectral approaches to the assessment of coupling and causality (Chicharro, 2011;Faes et al, 2021), yielding time domain measures of total coupling, Granger causality and instantaneous interaction that can be retrieved by integrating the corresponding spectral measures over the whole frequency axis. Moreover, the framework is formulated in a way such that two alternative ways are provided to treat instantaneous effects in both time and frequency domains.…”