Key points are not available for this paper at this time.
Oxford Medical has introduced an automatic sleep stager based on the stage-scoring criteria by Rechtschaffen and Kales. With our study we intended to examine whether the results of the stager (version 3.0) match those of the visual evaluation by two independent raters. We also wanted to test the reliability of this automatic sleep stage-scoring system. Ten somnopolygrams of subjects without sleep disturbances served as a basis for the comparison. Each sleep recording was scored twice automatically by the stager, twice visually by the first rater, and once by the second rater. The two automatic analyses of the somnopolygrams differed by 4.3% in a total of 13,850 epochs (1 epoch delta 20 s) regarding sleep stage scoring. The difference between the first and the second visual evaluation by the same rater amounted to 5.7%, whereas the results of the two independent raters deviated by 8.7%. Compared with the results of the visual analysis reached as a consensus by both raters--the so-called optimized visual analysis--the stager showed a 26.9% difference. The automatic analysis scored fewer epochs as stages wake, rapid eye movement (REM), and 2 and more as stages 1, 3, and 4. The sleep stager's frequent difficulty in identifying stage wake correctly as well as its incorrect allocation to other stages--mainly stage REM--could lead to misinterpretations of sleep recordings, whereas the increase in stages 1, 3, and 4, as compared with visual scoring, was negligible.
Kubicki et al. (Wed,) studied this question.