[Sleep in the young adult studied through automatic analysis of recordings]

Our laboratory sleep polygraphic recordings are all automatically scored by a scoring system previously described. The automatic scoring reliability has already been estimated by a comparison with visual scoring. Nevertheless, it was interesting to investigate whether the picture of sleep given by this system, of a precisely defined group of subjects, was very different from that obtained by visual scoring for similar subjects in other laboratories. Therefore, we compared the results given by automatic analysis in a group of normal subjects aged 20--29 years with the results of Williams et al. (1974) for subjects of the same age. We thus mixed on purpose three sources of difference: the studied population, the laboratory and its habits, and the technique of scoring the records. This comparison showed a clear difference in the total time of sleep for men and women. This time, longer in our conditions, depended certainly on subjects and on recording conditions more than on automatic scoring. Sleep and stage latencies were not significantly different by the two scoring techniques. A difference appeared for waking latency, longer in our findings, but probably resulting from the definition of this parameter. The number of sleep cycles and their average duration were not different in the two techniques. Concerning sleep stages, stage 3 was slightly overestimated and stage 4 underestimated by automatic scoring compared with visual scoring, while there was no marked difference for the other stages. Finally, a larger number of stages are detected by the machine than by the human eye, which is probably due to the fact that the latter neglects some transitions from one stage to another when they are not important with regard to the context. Thus, apart from some differences resulting from the definition of parameters, there is a very large analogy between visual and automatic results of sleep scoring. The only effects of the latter are a slightly different assessment of the ratio of stages 3 and 4, in favour of stage 3, a larger estimate of the number of sleep stages, and a slight underestimation of REM sleep.