The trilinear decomposition method also has been used for resting EEG, to estimate spectral and spatial components. These trilinear components of the resting EEG have been used in a COGA study to reduce multiple testing of electrodes and frequency bands, where significant linkage/linkage disequilibrium and association was found between a trilinear beta EEG phenotype and GABRA2, a GABAA receptor gene, later found to be also associated with alcoholism (Edenberg et al. 2004; Porjesz et al. 2002a). (See the section “Electrophysiological Measures as Endophenotypes for Alcoholism.”) Trilinear decomposition also has been applied in several studies with EEG (Martinez-Montes et al. 2004; Miwakeichi et al. 2004) and EROs (Morup et al. 2006, 2008). Recently, Verleger and colleagues (2013) applied trilinear decomposition to understand the relationship between CNV and the P3 complex in a Go/No-Go paradigm and obtained relevant components. Trilinear decomposition also has been successfully applied to seizure localization and found to be more sensitive than visual interpretation of the EEGs recorded during a seizure (De Vos et al. 2007). Trilinear modeling has great utility in alcoholism, and further studies are currently being conducted.
