Computer Simulation of Sleep EEG Patterns with a Markov Chain Model

Abstract

For the most part, recent investigations of sleep and dream states of human subjects have followed a procedure which calls for the analysis of continuous electroencephalographic (EEG) recordings made during an all-night sleep on the part of various populations of subjects [1–5]. Comparisons of sleep EEG data from different investigations of human subjects have been found to be exceedingly difficult to achieve, in part because of the very large number of EEG tracings recorded during a single night of continuous sleep. For example, the analysis of one night’s sleep EEG recording may involve the rather formidable task of reducing as much as one third of a mile of EEG tracings into a meaningful form. Furthermore, the complete lack of standardization of methods of data presentation in sleep research renders comparisons among the results of sleep studies published by different authors exceedingly difficult. Therefore, it is not surprising to find that existing methods of data reduction and reporting such as histograms, tables showing percentage of time spent in different stages of sleep, and all-night sleep graphs are all either completely inadequate or lead to gross oversimplifications of results and do not facilitate easy comparisons. In addition, none of the existing methods of analyzing all-night sleep EEG patterns lend themselves to the analysis of the observed phenomenon of continuous shifts in the stages of sleep.