A Multidimensional Functional Principal Components Analysis of EEG Data

• Activity Number: 471
• Type: Invited
• Date/Time: Wednesday, August 3, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Mental Health Statistics Section
• Author(s): Damla Senturk* and Kyle Hasenstab and Aaron Scheffler and Donatello Telesca and Catherine Sugar and Shafali Jeste
• Companies: University of California at Los Angeles and University of California at Los Angeles and University of California at Los Angeles and University of California at Los Angeles and University of California at Los Angeles and University of California at Los Angeles
• Keywords: event-related potentials data ; multilevel functional principal component analysis ; electroencephalography

Abstract

The data created in typical event-related potentials (ERP) studies is rich and multidimensional. Each stimulus, referred to as a trial, results in an ERP function with paradigm specific phasic components. Hence the experiment creates functional data (ERP functions) for each subject, collected longitudinally over trials at multiple electrodes placed on the scalp. We propose a multilevel longitudinal functional principal components decomposition for EEG data, which we refer to as the multi-dimensional functional principal components analysis (MD-FPCA), embodying all three dimensions (functional, longitudinal and spatial) of the EEG data. The proposed decomposition parallels the full complexity of the data, without stringent assumptions or data reductions along any dimension. We demonstrate the methodology by applications to a visual implicit learning study on young children with Autism Spectrum Disorder.