Graphical Model and Model Search for Medical Data Analysis

Learning the electrophysiological activities inside the human mind is a significant step toward studying the human brain. Systems, such as electroencephalography, are significant instruments for considering the neurophysiologic activities, in view of their high value of temporal and spatial resolution. In the biomedical research, identifying brain abnormalities such as autism spectrum disorder through electroencephalography (EEG) signals is an extremely exhausting issue for specialists and human services experts. The high volume of data available with EEG will be a useful biomarker for the classification of autism and typical children. Traditional techniques face challenges to deal with such big data. So we present a strategy for autism identification by analyzing the EEG signal through mathematical model. One such modeling using graph theory is applied in this work. The EEG signals are acquired from 3 autism and 3 typical children. The functional connectivity among the neuron regions are plotted through small world networks. From this graphical models using a software tool Gephi, the graphical parameters as betweenness centrality, degree, weighted degree, closeness centrality, modularity, and clustering coefficient are calculated. There is significant difference among these parameters between autistic and typical children.