EEG Power Spectrum as a Biomarker of Autism: A Pilot Study

Childhood autism is one of the disorders within the Autism Spectrum Disorder (ASD) that includes, among others, Asperger’s syndrome and Rett syndrome. These are collectively termed spectrum disorders due to the variability in symptom presentation. The present study aimed to assess whether Power Spectrum estimates of Electroencephalogram (EEG) can serve as a biomarker for autism. EEG data were collected from ASD and control participants during a short-memory task and pre-processed to remove noise and artifacts. Subsequently, Power Spectral Density (PSD) estimates were obtained using the modified covariance method. These estimates were then analyzed using the Kruskal-Wallis test to determine if there were statistically significant differences between autistic and control subjects. The results confirmed statistical differences in PSD estimates, which were further classified using the k-nearest neighbor (knn) algorithm, achieving an accuracy of 89.29%. This indicates that the EEG of autistic and control individuals contains distinguishable features, suggesting that EEG power spectrum could be a potential biomarker for autism.