PaulMuadDib

Thanks Simon. I was just wondering if the square brackets (a list) where syntaxically the correct way to retain several (available) channels like 'EEG(sec)' and 'EEG' (which presumably are extracted...

I thought the scorer could be used with the two available EEG channels of the PSG (plus EOG & EMG, but data is shaped with a last dim of 3...

Sorry for bringing confusion here. I am indeed reusing the script run_sample.py, where one EEG channel (among the two available), one EMG and one EOG channels are used. My first...

I think this for-loop is precisely the one which picks up 'EEG(SEC)' or 'EEG2' (when the 'EEG' dict. item is declared as a list of strings like in: `channels =...

Or, if you want to retain the possibility to select ('EEG(SEC)' OR 'EEG2') AND 'EEG' like with: ` channels = {'EEG': (['EEG(SEC)','EEG2], 'EEG'), 'EMG': 'EMG', 'EOG': 'EOG(L)'} ` (useful for...

Btw, the best model for the 50 males between 38 and 58 yrs drawn from the 200 cases of SHH1 only reaches: **fold 6: val acc/f1: 0.80947/0.73136, test acc/f1: 0.76639/0.68004**...

I was not aware that the dynamical range of the EEG channels was limited to 120µV. Last week, I used Spinky to see whether it could produce useful features for...

Using a balanced training, or not, made a difference... would it make sense to you the more I select similar individuals (gender, age, bmi, %rem, etc...) in this database, the...

Yes Simon, thank you for this reference, the quality of the human scoring on which the automatic scorer is training maybe a main driver for this... but difficult to assess...

The EDFx sleep database is the one scored with the largest number (4 of them) of human experts ?