A Comparison of Doubly Hierarchical Discriminant Analyses for Multiple Class Longitudinal Data from EEG Experiments

Abstract

This paper proposes a general and simple procedure that can be applied to establish classification rules for application to multiple-class longitudinal data. The procedure is applied to preclinical pharmaco-electroencephalogram (EEG) studies aiming at characterizing psychotropic drug effects on the basis of spectral EEG analysis. It is a flexible hierarchical supervised learning tool that takes into account the specific nature of the multiple drug classes, as well as the longitudinal aspect of the data. Several variations of this procedure are applied to the EEG data, generally producing comparable results, in particular similar association between the sleeping stages and the psychotropic drug classes.