A reproducible pipeline for simulating individual and inter-brain neural dynamics from structural connectivity
Annemarie Wolff1, Mahta Ramezanian Panahi2, Imme Christina Schultheiß3
1C.H.U. Sainte-Justine Azrieli Research Centre, Université de Montréal, Montréal, Québec, Canada.
Background:
Computational models can link anatomical structure to neural dynamics. However, there is no validated, reproducible pipeline that simulates individual and inter-brain EEG-like dynamics from an individual's own structural connectome.
New Method:
To address this, we developed a comprehensive computational pipeline that combines probabilistic tractography from Diffusion Magnetic Resonance Imaging (dMRI) with dynamical systems modeling to simulate EEG-like neural dynamics from individual connectomes. It extracts anatomical connectivity, intra-individual functional connectivity, and inter-individual (two-brain) functional coupling features. It then evaluates their group-discriminative information with random-forest classification under repeated nested cross-validation. Simulated phase-locking values are calibrated against empirical resting-state EEG.
Results:
In a test sample of individuals with autism spectrum disorder (ASD) and matched controls (16 each), simulated intra-individual synchrony added classification information beyond anatomy, raising accuracy from 71.2% (AUC 0.843) to 77.6% (AUC 0.871), in a sample where groups did not differ in mean structural connectivity.
Comparison With Existing Methods:
Unlike current single-brain simulations, this pipeline generates and jointly simulates two individual brain models within a single system, and reports the contribution of each feature level explicitly.
Conclusions:
The pipeline provides a reproducible, extensible framework for deriving informative individual-level dynamic features from anatomy. The two-brain extension generates and ranks candidate inter-brain features, providing a principled way to identify between-brain measures for hypothesis-driven testing.
