Covariate-adjusted hybrid principal components analysis for region-referenced functional EEG data

Area of Science:

• Neuroscience
• Biostatistics
• Developmental Psychology

Background:

• Electroencephalography (EEG) data provides region-referenced functional insights into neurodevelopment.
• Analyzing high-dimensional EEG data requires methods to capture complex dependencies and age-related variations.
• Autism Spectrum Disorder (ASD) diagnosis can be informed by neurodevelopmental trajectory differences.

Purpose of the Study:

• To develop a statistical method for analyzing EEG spectral variation across development and scalp regions.
• To identify neurodevelopmental differences between typically developing (TD) children and children with ASD using EEG data.
• To account for covariate-dependent heteroscedasticity in EEG signal analysis.

Main Methods:

• Proposed a covariate-adjusted hybrid principal components analysis (CA-HPCA) for EEG data.
• CA-HPCA integrates vector and functional principal components analysis.
• The method assumes weak separability of the covariance process conditional on covariates for efficient estimation.

Main Results:

• The study models patterns of alpha spectral variation, not just peak frequency.
• CA-HPCA allows for stable and computationally efficient estimation of EEG data.
• Novel insights into neurodevelopmental variations between TD and ASD children were obtained.

Conclusions:

• CA-HPCA offers a robust approach for analyzing complex, covariate-dependent EEG data.
• The methodology enhances understanding of neurodevelopmental trajectories in relation to ASD.
• This approach can reveal subtle differences in brain function related to neurodevelopmental conditions.