WITHDRAWN: Functional Mixed-Effects Modeling of Electroencephalographic (EEG) Data

• Activity Number: 256
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Shubhankar Ray
• Companies: Merck & Co., Inc.
• Keywords: EEG power spectra ; functional data analysis

Abstract

Many neurological disorders show unique power spectral (PSD) profiles which makes them popular in medicine. We want to estimate the drug effects relative to placebo and its interactions with other factors.

Although, semiparametric functional mixed-effects models can be used, fitting with an adequate covariance structure is often computationally prohibitive. Fan and Zhang (1999) propose a fast two-step estimation procedure that does not make any assumptions about the error covariance.

We compare these two approaches and propose a simplification of the former, where the observed PSDs are first projected onto a spline basis. Empirical spline coefficients are later used as observations in a mixed-effects model. This reduces the computational burden but also allows flexibility with error-covariance structures. We provide results of the comparison for both simulated and real EEG datasets.