Characterizing Autism Spectrum Disorder Through Fusion of Local Cortical Activation and Global Functional...

Area of Science:

• Neuroscience
• Developmental Psychology
• Biomedical Engineering

Background:

• Social interaction deficits are a core feature of autism spectrum disorder (ASD), influenced by personal experiences and social environments.
• Neuroimaging studies have established a link between social impairments and altered brain activity in individuals with ASD.
• Current behavioral assessments for ASD can be subjective and lack precise accuracy.

Purpose of the Study:

• To develop and validate a novel method for assessing and identifying ASD using a social cognitive game-based paradigm combined with electroencephalography (EEG) signaling features.
• To investigate neurophysiological differences in brain activity and functional connectivity between typically developing (TD) individuals and autistic preadolescents and teenagers during social interaction tasks.
• To evaluate the efficacy of machine learning models in characterizing ASD based on EEG-derived neurophysiological features.

Main Methods:

• Recruited typically developing (TD) participants and autistic preadolescents/teenagers for a social cognitive game.
• Recorded 12-channel electroencephalography (EEG) signals during the game, followed by preprocessing to analyze local brain activities (ERPs, time-frequency features) and global functional connectivity (PLIs).
• Employed machine learning models, specifically support-vector machines, to assess neurophysiological features for ASD characterization.

Main Results:

• Autistic individuals showed distinct event-related potential (ERP) differences, including lower late positive potential (LPP) amplitudes and larger P200 amplitudes in parietal regions compared to TD participants.
• Reduced theta synchronization and aberrant functional connectivity patterns were observed in the ASD group during social tasks.
• Machine learning models utilizing ERP and brain oscillation features achieved high performance metrics (100% sensitivity, 91.7% specificity, 95.8% accuracy) for ASD characterization.

Conclusions:

• Social interaction difficulties in ASD are associated with specific, quantifiable patterns of brain activation and connectivity.
• The proposed social training interface and EEG-based neurophysiological features show significant potential as an objective and accurate cognitive assessment tool for ASD.
• This approach offers a promising alternative to traditional, potentially subjective, behavioral assessments for identifying and understanding ASD.