نهمین کنگره نقشه برداری مغز

نمایش مقاله

Title:

Virtual Brain Twins: Mathematical Foundations, Computational Modeling, and Comparative Simulation of Neurological and Psychiatric Disorders

Subject:

Artificial Intelligence and Brain Mapping

Presenter Author:
Name:	Mohammad Fayaz

Author:
Name 1:	Mohammad Fayaz - Corresponding - Presenter
Affiliation 1:	Assistant Professor of Biostatistics, School of Medicine, Shahed University, Tehran, Iran. ORCID: 0000-0002-5643-9763

Name 2:
Affiliation 2:

Name 3:
Affiliation 3:

Name 4:
Affiliation 4:

Name 5:
Affiliation 5:

Name 6:
Affiliation 6:

Name 7:
Affiliation 7:

Name 8:
Affiliation 8:

Name 9:
Affiliation 9:

Name 10:
Affiliation 10:

Abstract

Original Research

introduction :

Virtual brain twins are an emerging computational framework for personalized modeling of brain dynamics through the integration of subject-specific neuroimaging, mathematical brain network models, and numerical simulations. Rather than reproducing the brain at complete biological resolution, these models seek a balance between biological realism, mathematical abstraction, and computational tractability. Virtual brain approaches have been investigated in neurological and psychiatric disorders, including epilepsy, Alzheimer’s disease, Parkinson’s disease, multiple sclerosis, and schizophrenia, with potential applications in understanding disease mechanisms, identifying biomarkers, and evaluating therapeutic interventions.

Methods:

This study aims to identify, characterize, implement, and comparatively evaluate virtual brain and digital brain-twin models used for neurological and psychiatric disorders. The underlying framework is based primarily on brain network models (BNMs), in which brain regions are represented using neural mass, neural field, or nonlinear oscillator models and interconnected according to structural connectivity, typically derived from diffusion MRI. Coupling functions determine interactions among brain regions, while conduction velocities and time delays can represent signal transmission along white-matter pathways. The resulting differential equations are numerically integrated to generate whole-brain dynamics. Forward models can subsequently transform simulated neural activity into signals comparable with empirical measurements, including EEG, MEG, intracranial EEG (sEEG), and fMRI. As a computational demonstration, a representative model was implemented in Python using The Virtual Brain (TVB), combining subject-specific structural connectivity with nonlinear local dynamics. Simulated activity was examined in both time-domain and temporally aggregated representations. Model structure, personalization strategies, parameter estimation, computational implementation, and validation approaches were considered within a comparative framework.

Results:

The computational demonstration illustrates how structural connectivity, local nonlinear dynamics, coupling strength, and conduction delays jointly shape emergent whole-brain activity. The framework also demonstrates how model parameters can be modified to represent disease-related alterations and how subject-specific data can support model personalization. Across disorders, different modeling strategies target distinct pathophysiological mechanisms. In epilepsy, subject-specific models can characterize abnormal propagation and estimate epileptogenic regions. In Alzheimer’s disease, models can investigate alterations in oscillatory activity and large-scale network integration. In multiple sclerosis, changes in structural connectivity and conduction properties can be incorporated to examine their effects on brain dynamics. In Parkinson’s disease, cortico-basal-ganglia-thalamocortical models provide a framework for investigating pathological dynamics and neuromodulatory interventions. In schizophrenia, whole-brain models can be used to investigate altered excitation–inhibition balance and abnormal structure–function relationships. Bayesian inference, optimization, and machine learning can further support parameter estimation and uncertainty quantification.

Conclusion:

Virtual brain twins provide a flexible framework for linking individual brain anatomy to large-scale dynamics and disease-related alterations. Comparative evaluation of model structures, personalization methods, data requirements, validation strategies, and uncertainty quantification can clarify the strengths and limitations of existing approaches and support the development of more reproducible and clinically relevant personalized brain models.

Keywords:

Virtual Brain Twins, Brain Network Models, Personalized Brain Modeling, Computational Neuroscience, Neurological and Psychiatric Disorders