ARMA modeling for the diagnosis of controlled epileptic activity in young childrenARMA modeling for the diagnosis of controlled epileptic activity in young children Πλήρης Δημοσίευση σε Συνέδριο Conference Full Paper 2015-10-252008enParametric models are widely used for EEG data analysis. In this experimental study an autoregressive moving average (ARMA) model was used to extract spectral features within defined frequency bands which were then used to discriminate a group of children with controlled mild epilepsy from an age- and sex-matched control group. This study differs from other published works in that it shows that this technique can be used as a biomarker to distinguish the epileptic subjects specifically when the EEG recordings of these subjects are clinically diagnosed as normal. Using the spectral features and a linear discriminant classifier a global classification score of up to 85% was achieved on our clinical data. Furthermore the results showed that epileptic children have significantly higher spectral power in frequency bands up to 45 Hz, with the largest difference occurring within the alpha band.http://creativecommons.org/licenses/by/4.0/25-30 3rd International Symposium on Communications, Control and Signal Processing Zervakis Michalis Ζερβακης Μιχαλης Camilleri K. P. Μιχελογιάννης Σήφης Michelogiannis Sifis Micheloyannis Sifis Fabri S. G. Cassar T. A. Institute of Electrical and Electronics Engineers Epilepsy Brain modeling Autoregressive processes EEG Electroencephalography Frequency Parametric statistics Data analysis Data mining Feature extraction Biomarkers