Three different methods of automated sleep staging are described and c
ompared in the paper. The interesting aspect of the comparison is that
the inputs to the three different information processing models (an e
xpert system, a belief automaton, and a neural network) are the output
s of the same frontend processor that extracts EEG features. We found
out that the level of man-machine agreement is very similar among the
systems (around 85%) approaching the level of agreement required among
human experts. However, the similarity of performance in such a diver
sified set of approaches points out the need for enhancing the accurac
y of the first level of analysis (EEG feature extraction).