Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

nCREANN:Nonlinear Brain Connectivity and Diverse Applications

Nasibeh Talebi¹, Christian Beste; ¹Technische Universität Dresden

Presenter: Nasibeh Talebi

Understanding directed brain connectivity is crucial in neuroscience, yet traditional linear connectivity methods may oversimplify neural interactions. We introduce nCREANN, nonlinear causal relationship estimation by artificial neural networks, which models the neural dynamics by a nonlinear multivariate autoregressive (nMVAR) process and estimates directed connectivity. This method leverages the Taylor expansion of nonlinear input-output mapping of the neural network to dissociate linear and nonlinear connectivity patterns. We summarize considerations and diverse applications of nCREANN in neuroscience studies. Results highlight distinct linear and nonlinear connectivity patterns in Autism Spectrum Disorder (ASD) and Attention-Deficit/Hyperactivity Disorder (ADHD) subjects, a superior classification accuracy of ADHDs (up to 99%), and deeper insights into neural mechanisms underlying adaptive behavior, event segmentation, metacontrol processes, and dynamic working memory gating. nCREANN provides a powerful tool for uncovering nuanced brain dynamics and enhancing understanding of neural disorders.

Topic Area: Brain Networks & Neural Dynamics

Extended Abstract: Full Text PDF