Introduction to information theory and network reconstruction

• Grant number: 25/21841-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2026
• End date: January 31, 2027
• Field of knowledge: Physical Sciences and Mathematics - Physics
• Principal Investigator: Thomas Kaue Dal Maso Peron
• Grantee: Pietra Gullo Salgado Chaves
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

Functional networks have been widely used in studies on epilepsy, autism, Alzheimer's, and several other neurological conditions, as they allow us to characterize the connectivity between different brain regions. From these networks, metrics are extracted that reveal patterns of interaction that distinguish patients from control groups. However, one of the main challenges in this field is choosing the most appropriate connectivity metric to quantify the relationships between brain regions, since different measures can capture different aspects of neural dynamics.In this context, this project proposes a comparative approach to methods for reconstructing functional networks from electroencephalogram (EEG) data. The methodology involves generating synthetic networks, created randomly and used as a benchmark for evaluating the measurements. A dynamic signal is then assigned to each node in the network, simulating its temporal evolution using chaotic models, such as Rössler's. Different connectivity metrics will be applied to these signals, including both classical information theory measures and metrics widely explored in the literature: Pearson correlation, mutual information, nonlinear interdependence measures, and phase synchronization. Each of these will generate functional networks that will subsequently be compared to the original reference network using similarity indices, such as the Jaccard coefficient or related metrics.As an expected result, the research seeks to identify which measures are most effective in reconstructing functional networks, considering the nature of the signal and the interactions involved. Thus,the aim is to provide support for choosing the most appropriate metrics in studies of brain conditions,contributing to more accurate analyses and a better understanding of neural connectivity. (AU)