Related Experiment Video
Updated: Jun 5, 2025
08:51
Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms
Published on: November 1, 2019
5.6K
Multiscale effective connectivity analysis of brain activity using neural ordinary differential equations
Yin-Jui Chang1, Yuan-I Chen1, Hannah M Stealey1
1Biomedical Engineering, University of Texas at Austin, Austin, TX, United States of America.
Plos One
|December 4, 2024
Summary
A new deep learning model, msDyNODE, captures multiscale brain activity and communication across different brain regions. This model reveals behavior-dependent causal interactions, advancing mechanistic studies of neural processes.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Deep Learning
Background:
- Understanding brain region communication requires models of neural dynamics across multiple scales.
- Current multimodal brain activity measurements lack comprehensive multiscale dynamical models.
Purpose of the Study:
- To introduce a neurobiological-driven deep learning model, msDyNODE, for describing multiscale brain communications.
- To model collective neural activity governing cognition and behavior.
Main Methods:
- Developed a deep learning model named multiscale neural dynamics neural ordinary differential equation (msDyNODE).
- Validated msDyNODE using computational simulations and electrophysiological experiments.
Main Results:
- msDyNODE successfully captured multiscale neural activity.
- Derived causal interactions aligned with mammalian central nervous system neuroanatomy.
- Identified behavior-dependent neural interactions.
Conclusions:
- msDyNODE provides a novel approach for mechanistic multiscale studies of neural processes.
- The model effectively describes multiscale brain communications and their link to behavior.
