I am currently screening the datasets, looking for ones that contain re-scored events, ideally SDBs and not sleep stages, made by 2 technicians or 1 technician twice. I want to assess how trustworthy the annotations are for machine learning. I searched the datasets and had trouble locating one that would meet this criteria, but did I miss any?
Hey - thanks for checking out the resource. Unfortunately, I don't think you will find any datasets on our site that meet these criteria. Many of the studies/trials routinely included scorer reliability exercises, but these were done strictly for internal QA purposes.
Hi thanks for confirmation.