Bayesian inference of high-dimensional, cluster-structured ordinary differential equation models with applications to brain connectivity studies

Tingting Zhang, Qiannan Yin, Brian S Caffo, Yinge Sun, Dana Boatman-Reich

Research output: Contribution to journalArticle

Abstract

We build a new ordinary differential equation (ODE) model for the directional interaction, also called effective connectivity, among brain regions whose activities are measured by intracranial electrocorticography (ECoG) data. In contrast to existing ODE models that focus on effective connectivity among only a few large anatomic brain regions and that rely on strong prior belief of the existence and strength of the connectivity, the proposed high-dimensional ODE model, motivated by statistical considerations, can be used to explore connectivity among multiple small brain regions. The new model, called the modular and indicator-based dynamic directional model (MIDDM), features a cluster structure, which consists of modules of densely connected brain regions, and uses indicators to differentiate significant and void directional interactions among brain regions. We develop a unified Bayesian framework to quantify uncertainty in the assumed ODE model, identify clusters, select strongly connected brain regions, and make statistical comparison between brain networks across different experimental trials. The prior distributions in the Bayesian model for MIDDM parameters are carefully designed such that the ensuing joint posterior distributions for ODE state functions and the MIDDM parameters have well-defined and easy-to-simulate posterior conditional distributions. To further speed up the posterior simulation, we employ parallel computing schemes in Markov chain Monte Carlo steps. We show that the proposed Bayesian approach outperforms an existing optimization-based ODE estimation method. We apply the proposed method to an auditory electrocorticography dataset and evaluate brain auditory network changes across trials and different auditory stimuli.

Original languageEnglish (US)
Pages (from-to)868-897
Number of pages30
JournalAnnals of Applied Statistics
Volume11
Issue number2
DOIs
StatePublished - Jun 1 2017

Fingerprint

Bayesian inference
Ordinary differential equations
Brain
Connectivity
Ordinary differential equation
High-dimensional
Posterior distribution
Model
Feature Model
Differential equations
Bayesian Model
Parallel Computing
Conditional Distribution
Voids
Prior distribution
Markov Chain Monte Carlo
Differentiate
Interaction
Bayesian Approach
Joint Distribution

Keywords

  • Bayesian inference
  • Cluster structure
  • Directional brain networks
  • Network edge selection
  • ODE models

ASJC Scopus subject areas

  • Statistics and Probability
  • Modeling and Simulation
  • Statistics, Probability and Uncertainty

Cite this

Bayesian inference of high-dimensional, cluster-structured ordinary differential equation models with applications to brain connectivity studies. / Zhang, Tingting; Yin, Qiannan; Caffo, Brian S; Sun, Yinge; Boatman-Reich, Dana.

In: Annals of Applied Statistics, Vol. 11, No. 2, 01.06.2017, p. 868-897.

Research output: Contribution to journalArticle

@article{01e6d7f9ca1c4011bdce8bea1df50888,
title = "Bayesian inference of high-dimensional, cluster-structured ordinary differential equation models with applications to brain connectivity studies",
abstract = "We build a new ordinary differential equation (ODE) model for the directional interaction, also called effective connectivity, among brain regions whose activities are measured by intracranial electrocorticography (ECoG) data. In contrast to existing ODE models that focus on effective connectivity among only a few large anatomic brain regions and that rely on strong prior belief of the existence and strength of the connectivity, the proposed high-dimensional ODE model, motivated by statistical considerations, can be used to explore connectivity among multiple small brain regions. The new model, called the modular and indicator-based dynamic directional model (MIDDM), features a cluster structure, which consists of modules of densely connected brain regions, and uses indicators to differentiate significant and void directional interactions among brain regions. We develop a unified Bayesian framework to quantify uncertainty in the assumed ODE model, identify clusters, select strongly connected brain regions, and make statistical comparison between brain networks across different experimental trials. The prior distributions in the Bayesian model for MIDDM parameters are carefully designed such that the ensuing joint posterior distributions for ODE state functions and the MIDDM parameters have well-defined and easy-to-simulate posterior conditional distributions. To further speed up the posterior simulation, we employ parallel computing schemes in Markov chain Monte Carlo steps. We show that the proposed Bayesian approach outperforms an existing optimization-based ODE estimation method. We apply the proposed method to an auditory electrocorticography dataset and evaluate brain auditory network changes across trials and different auditory stimuli.",
keywords = "Bayesian inference, Cluster structure, Directional brain networks, Network edge selection, ODE models",
author = "Tingting Zhang and Qiannan Yin and Caffo, {Brian S} and Yinge Sun and Dana Boatman-Reich",
year = "2017",
month = "6",
day = "1",
doi = "10.1214/17-AOAS1021",
language = "English (US)",
volume = "11",
pages = "868--897",
journal = "Annals of Applied Statistics",
issn = "1932-6157",
publisher = "Institute of Mathematical Statistics",
number = "2",

}

TY - JOUR

T1 - Bayesian inference of high-dimensional, cluster-structured ordinary differential equation models with applications to brain connectivity studies

AU - Zhang, Tingting

AU - Yin, Qiannan

AU - Caffo, Brian S

AU - Sun, Yinge

AU - Boatman-Reich, Dana

PY - 2017/6/1

Y1 - 2017/6/1

N2 - We build a new ordinary differential equation (ODE) model for the directional interaction, also called effective connectivity, among brain regions whose activities are measured by intracranial electrocorticography (ECoG) data. In contrast to existing ODE models that focus on effective connectivity among only a few large anatomic brain regions and that rely on strong prior belief of the existence and strength of the connectivity, the proposed high-dimensional ODE model, motivated by statistical considerations, can be used to explore connectivity among multiple small brain regions. The new model, called the modular and indicator-based dynamic directional model (MIDDM), features a cluster structure, which consists of modules of densely connected brain regions, and uses indicators to differentiate significant and void directional interactions among brain regions. We develop a unified Bayesian framework to quantify uncertainty in the assumed ODE model, identify clusters, select strongly connected brain regions, and make statistical comparison between brain networks across different experimental trials. The prior distributions in the Bayesian model for MIDDM parameters are carefully designed such that the ensuing joint posterior distributions for ODE state functions and the MIDDM parameters have well-defined and easy-to-simulate posterior conditional distributions. To further speed up the posterior simulation, we employ parallel computing schemes in Markov chain Monte Carlo steps. We show that the proposed Bayesian approach outperforms an existing optimization-based ODE estimation method. We apply the proposed method to an auditory electrocorticography dataset and evaluate brain auditory network changes across trials and different auditory stimuli.

AB - We build a new ordinary differential equation (ODE) model for the directional interaction, also called effective connectivity, among brain regions whose activities are measured by intracranial electrocorticography (ECoG) data. In contrast to existing ODE models that focus on effective connectivity among only a few large anatomic brain regions and that rely on strong prior belief of the existence and strength of the connectivity, the proposed high-dimensional ODE model, motivated by statistical considerations, can be used to explore connectivity among multiple small brain regions. The new model, called the modular and indicator-based dynamic directional model (MIDDM), features a cluster structure, which consists of modules of densely connected brain regions, and uses indicators to differentiate significant and void directional interactions among brain regions. We develop a unified Bayesian framework to quantify uncertainty in the assumed ODE model, identify clusters, select strongly connected brain regions, and make statistical comparison between brain networks across different experimental trials. The prior distributions in the Bayesian model for MIDDM parameters are carefully designed such that the ensuing joint posterior distributions for ODE state functions and the MIDDM parameters have well-defined and easy-to-simulate posterior conditional distributions. To further speed up the posterior simulation, we employ parallel computing schemes in Markov chain Monte Carlo steps. We show that the proposed Bayesian approach outperforms an existing optimization-based ODE estimation method. We apply the proposed method to an auditory electrocorticography dataset and evaluate brain auditory network changes across trials and different auditory stimuli.

KW - Bayesian inference

KW - Cluster structure

KW - Directional brain networks

KW - Network edge selection

KW - ODE models

UR - http://www.scopus.com/inward/record.url?scp=85026286979&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85026286979&partnerID=8YFLogxK

U2 - 10.1214/17-AOAS1021

DO - 10.1214/17-AOAS1021

M3 - Article

AN - SCOPUS:85026286979

VL - 11

SP - 868

EP - 897

JO - Annals of Applied Statistics

JF - Annals of Applied Statistics

SN - 1932-6157

IS - 2

ER -