Dear Clinicians/Researchers working on sleep disorders,

We are working on the automated detection of sleep disorders using EEG. While we can get good accuracy with binary classifiers distinguishing normal subjects from any one particular sleep disorder, the obtained accuracy is very low, when we attempt multiclass classifiers, where the classifier is expected to identify a specific sleep disorder from the following list: insomnia, narcolepsy, REM behavior disorder, bruxism, nocturnal frontal lobe epilepsy, periodic leg movement syndrome, and sleep-disordered breathing. I would like to know from the clinicians (i) what features they look for in identifying the above disorders (differential diagnosis) (ii) is EEG a meaningful signal to use for this purpose. if not, what other signals could be used? (iii) how is a clinical diagnosis performed?

I shall be very grateful for any useful input in this direction.

Best regards.

Ram

More A.G. Ramakrishnan's questions See All
Similar questions and discussions