Brain Network Constancy and Participant Recognition: an Integrated Approach to Big Data and Complex Network Analysis

Area of Science:

• Neuroscience
• Big Data Analytics
• Cognitive Science

Background:

• Electroencephalogram (EEG) data is increasingly utilized for studying human cognitive behavior, driven by big data sharing and standardization.
• Existing research primarily examines brain network topology changes during cognitive tasks, with limited exploration of multi-cognitive behavior stability and multi-participant network recognition.

Purpose of the Study:

• To investigate the steady-state characteristics of multi-cognitive behaviors and enable recognition of multi-participant brain networks using EEG data.
• To establish the concept of a 'brain fingerprint' based on the unique and stable spectral distribution of individual brain networks.

Main Methods:

• Utilized EEG data from 99 healthy participants from PhysioBank.
• Calculated symbolic transfer entropy (STE) between 64 electrode sequences.
• Constructed directed minimum spanning tree (DMST) brain networks and analyzed the eigenvalue spectrum of the STE matrix for each participant and cognitive state.

Main Results:

• Spectrum distributions of different cognitive states within the same participant showed relative stability.
• Spectrum distributions for the same cognitive state varied considerably across different participants, supporting the uniqueness of brain networks.
• A novel Spectral Distribution Set Scoring (SDSS) method was developed, achieving 69.35% accuracy in identifying participants based on their brain network spectral distributions.

Conclusions:

• The study provides strong evidence for the existence of unique 'brain fingerprints' in human brain networks.
• The developed SDSS method offers a novel approach for dynamic identification of individuals based on their brain network characteristics.