Abstract Mkid 8

The scoring system of Rechtschaffen and Kales (1968) is practically the only method of visual sleep analysis in spite of its serious drawbacks. The most important ones are the fixed, long epochs, the insufficient number of sleep stages and the ignorance of EEG topography. In this preliminary work a method to overcome these limitations in visual scoring is presented. Segments of variable length and more stage categories than in the standard system were used. The method was preliminarily applied to the analyses of multiple sleep latency tests (MSLT) from three subjects. All recordings were classified both by the standard method of Rechtschaffen and Kales (RKS, 1968) and by visual adaptive scoring (VAS). VAS was more sensitive to short sleep spells than RKS and gave a shorter and according to our opinion a more accurate sleep onset latency. The rating of two subjects changed from borderline to pathological. One became more normal because of different scoring of one segment. The percentages of stages obtained by VAS were only slightly different from those obtained by RKS and did not reach statistical significance. Yet, the infrastructure, short lapses of alertness, short sleep episodes and arousals, were more precisely revealed. Although VAS is sometimes more time consuming than traditional scoring it is a model that fits the sleep/wake process more closely than RKS and can therefore be regarded as a better model. The improved temporal resolution gives electrophysiologically more stationary epochs. This will be especially important with computer analysis. When used for scientific work VAS helps to provide more profound understanding of underlying mechanisms of sleep and arousal as well as their correlates. Sometimes VAS is also easier to apply because many of the problems with RKS are due to the epoch boundaries and landmarks not fitting the stage categories.