Conditional entropy approach to analyze cognitive dynamics in autism spectrum disorder

Area of Science:

• Neuroscience
• Cognitive Science
• Biomarkers

Background:

• Functional connectivity is documented in Autism Spectrum Disorder (ASD), but effective neural connectivity and information flow during complex tasks remain unclear.
• Understanding brain network dynamics under cognitive load is crucial for insights into ASD cognition.

Purpose of the Study:

• To investigate cognition-based neural dynamics and information exchange in the brain network of individuals with ASD under cognitive load.
• To identify potential biomarkers for cognitive status in ASD based on neural connectivity patterns.

Main Methods:

• Employed Conditional Entropy (CE) on task-activated Electroencephalogram (EEG) data from 22 individuals with ASD (8-18 years) and 18 Typically Developing (TD) individuals (6-17 years).
• Assessed causal influence between brain regions of interest (ROIs) during a task differentiating risky from neutral stimuli.
• Utilized Support Vector Machine (SVM) for classification of atypical information exchange.

Main Results:

• Higher CE in the frontal ROI and left hemisphere indicated atypical brain complexity in ASD.
• Poor Coupling Strength (CS) and lack of hemisphere lateralization correlated with lower cognition in ASD.
• CE accurately identified atypical information exchange (96.89% accuracy, AUC=0.987), suggesting alternative information pathways in ASD.

Conclusions:

• Significant alterations in information flow between ROIs were observed in ASD.
• Coupling Strength (CS) correlates with behavioral domains, potentially predicting cognitive decline in ASD.
• CS shows promise as a biomarker for identifying cognitive status in ASD with high discrimination.