Time-series analysis in EEG phase synchronization based on time-varying network modeling

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K19867
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 61030:Intelligent informatics-related
• Research Institution: Hiroshima University (2022); National Institute for Physiological Sciences (2020-2021)
• Principal Investigator: Yokoyama Hiroshi 広島大学, 統合生命科学研究科(理), 特任助教 (10829823)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 脳波 / 位相同期 / 脳機能ネットワーク解析 / ベイズ推定 / データ駆動型モデリング / 動的ネットワーク推定 / 変化点検知

Outline of Final Research Achievements

With the recent development of data science technology, although human neuroscience studies focused on the temporal dynamics of functional brain networks have attracted growing attention with various approaches, an effective and robust analytical methodology in a data-driven manner to interpret brain mechanisms behind measurement data has not been established. To address the issues, I proposed a new model-based approach to detect changes in dynamical network structures by combining the model-based network estimation with a phase-coupled oscillator model and change-point detection algorithm. To validate our proposed method, we applied it to mathematically modeled data and to empirical electroencephalogram (EEG) data. As a result, the method succeeded in detecting the change points of the dynamic brain networks in sub-second order based on measured EEG signals.

Free Research Field

脳情報工学

Academic Significance and Societal Importance of the Research Achievements

本研究では，データ駆動的なアプローチにて，観測脳波から脳機能ネットワークの時系列変化と検知と可視化の両方を同時に実現する手法を提案し，その妥当性を示すことができた．これらの結果は，脳機能ネットワークの過渡的な変化を時間分解能の高い脳波から本提案手法を基にミリ秒単位のスケールで検出できることを意味し，本提案手法を活用することで様々な認知課題実行時における脳のメカニズムとネットワークダイナミクスとの関係を定量的に議論することができる．脳のネットワーク動態の機能的役割が定量できれば，ヒトの認知機能評価に必要なサロゲートバイオマーカー探索などにも役立てることができ，医用工学的な応用が期待される．