Interpreting softmax predictions of sleep stage classification models
preprintposted on 06.10.2021, 19:40 by Iris HuijbenIris Huijben, Lieke WA Hermans, Allessandro C Rossi, Sebastiaan Overeem, Merel M van Gilst, Ruud JG van van Sloun
We used a dataset of nocturnal PSG recordings, collected as part of the Healthbed study, which main aim was development of technologies for sleep analyses. The dataset includes one clinical video-PSG recording for each subject, made according to the AASM recommendations in Sleep Medicine Center Kempenhaeghe Heeze, the Netherlands. The study included 96 (60 females) healthy subjects, with an age between 18 and 64. The exclusion criteria were: 1) any diagnosed sleep disorder, 2) a Pittsburgh Sleep Quality Index >= 6, or Insomnia Severity Index > 7, 3) indication of depression or anxiety disorder measured with the Hospital Anxiety and Depression Scale (score > 8), 4) pregnancy, shift work, use of any medication except for birth control medicine, and 5) presence of clinically relevant neurological or psychiatric disorders or other somatic disorders that could influence sleep.