Skip to main content
QUICK REVIEW

[论文解读] Measuring hierarchically-organized interactions in dynamic networks through spectral entropy rates: theory, estimation, and illustrative application to physiological networks

Laura Sparacino, Yuri Antonacci|arXiv (Cornell University)|Jan 20, 2024
Neural dynamics and brain functionNeuroscience被引用 3
一句话总结

本文提出了一种统一的信息-理论框架,利用谱熵率、互信息率和O-信息率,量化动态网络中分层组织的、频带特定的相互作用。该框架支持节点特异性、成对及高阶(3个及以上节点)相互作用的时间域与频域分析,揭示了在运动任务期间,α波段脑电图(EEG)动力学中存在显著冗余。

ABSTRACT

Recent advances in signal processing and information theory are boosting the development of new approaches for the data-driven modelling of complex network systems. In the fields of Network Physiology and Network Neuroscience where the signals of interest are often rich of oscillatory content, the spectral representation of network systems is essential to ascribe the analyzed interactions to specific oscillations with physiological meaning. In this context, the present work formalizes a coherent framework which integrates several information dynamics approaches to quantify node-specific, pairwise and higher-order interactions in network systems. The framework establishes a hierarchical organization of interactions of different order using measures of entropy rate, mutual information rate and O-information rate, to quantify respectively the dynamics of individual nodes, the links between pairs of nodes, and the redundant/synergistic hyperlinks between groups of nodes. All measures are formulated in the time domain, and then expanded to the spectral domain to obtain frequency-specific information. The practical computation of all measures is favored presenting a toolbox that implements their parametric and non-parametric estimation, and includes approaches to assess their statistical significance. The framework is illustrated first using theoretical examples where the properties of the measures are displayed in benchmark simulated network systems, and then applied to representative examples of multivariate time series in the context of Network Neuroscience and Network Physiology.

研究动机与目标

  • 开发一种一致且可扩展的框架,利用信息-理论度量量化动态网络中的节点特异性、成对及高阶相互作用。
  • 通过谱积分特性整合网络相互作用的时间域与频域表征,确保跨域一致性。
  • 通过将熵率、互信息率和O-信息率扩展至谱域,实现对生理相互作用的频带特定分析。
  • 提供一个实用工具箱,用于这些度量的参数化与非参数化估计,并包含显著性检验。
  • 展示该框架在捕捉真实生理网络中复杂、多尺度相互作用方面的实用性,特别是在网络神经科学与网络生理学中的应用。

提出的方法

  • 将熵率、互信息率(MIR)和O-信息率分别形式化为节点动力学、二元相互作用和多元(3个及以上节点)相互作用的核心度量。
  • 利用谱积分特性将所有度量扩展至谱域,实现信息动力学的频带特定分析。
  • 采用参数化与非参数化估计技术,并结合基于置换的假设检验评估统计显著性。
  • 将该框架应用于模拟网络系统以验证度量属性,并应用于真实多变量生理时间序列(如EEG、ECG、EMG)的运动任务实验数据。
  • 使用自 resampling 方法生成置换分布,以评估熵率、MIR 和 O-信息率谱特征的显著性。
  • 将所有度量在频带(如α、β)上进行整合,以总结时间域与频域中的整体信息动力学。
Figure 1: Information-theoretic and spectral representations of hierarchically-organized interactions in network systems. a) In Network Physiology, collective interactions among diverse organ systems (e.g., brain, heart, vascular, muscular and respiratory systems) are investigated recording biosigna
Figure 1: Information-theoretic and spectral representations of hierarchically-organized interactions in network systems. a) In Network Physiology, collective interactions among diverse organ systems (e.g., brain, heart, vascular, muscular and respiratory systems) are investigated recording biosigna

实验结果

研究问题

  • RQ1如何在时间域与频域之间一致地统一信息-理论度量,以量化动态网络中的分层相互作用?
  • RQ2谱分解在揭示多变量生理时间序列中具有生理意义的振荡相互作用方面发挥何种作用?
  • RQ3在运动任务期间,脑网络中的高阶相互作用(冗余与协同)在多大程度上占主导地位?它们在不同频带中如何变化?
  • RQ4所提出的框架能否检测到在仅考虑成对关系的网络模型中被掩盖的显著相互作用?
  • RQ5体积传导效应如何影响头皮EEG记录中冗余相互作用的检测?

主要发现

  • 谱熵率揭示了在α波段(约10 Hz)存在显著的振荡动力学,而β波段的贡献不显著,与已知的运动任务相关EEG模式一致。
  • 互信息率(MIR)显示大脑半球之间存在统计显著的成对耦合(X₁与X₂),其中α波段的值最高,表明存在强烈的半球间同步。
  • O-信息率分析在三节点相互作用中检测到显著的冗余贡献,尤其集中在α波段,表明在运动执行过程中,信息在脑区之间被冗余共享。
  • 整合的谱度量证实了在熵率与互信息率中,α波段动力学占主导地位,与运动相关振荡的既定神经生理学知识一致。
  • 该框架成功分离出频带特定的相互作用,揭示β波段活动未表现出显著耦合,凸显了谱滤波在检测生理相关动力学中的重要性。
  • 结果表明,体积传导可能对头皮EEG中观察到的冗余性有所贡献,因为已知α波段对体积传导效应敏感,而冗余相互作用在该频段尤为突出。
Figure 2: Venn-diagram representation of the information measures quantifying hierarchically-organized interactions in static networks of random variables. a) Mutual information between two random variables $V_{1}$ and $V_{2}$ , $I(V_{1};V_{2})$ , obtained as the difference between the entropy $H(V_
Figure 2: Venn-diagram representation of the information measures quantifying hierarchically-organized interactions in static networks of random variables. a) Mutual information between two random variables $V_{1}$ and $V_{2}$ , $I(V_{1};V_{2})$ , obtained as the difference between the entropy $H(V_

更好的研究,从现在开始

从阅读论文到最终审阅,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。