NSRR staff
Boston, MA
0000-0002-0506-8368
Top Topics
Recent Topics
Thanks for checking out the site. We (the NSRR team) haven't done much work with the STAGES dataset. MESA and CFS were centrally scored at an academic research center. STAGES, on the other hand, relied upon the clinical sleep scoring groups at each of the sites. Perhaps that explains some of the difference?
References:
You'll have to pull it from the cfs-visit5-dataset CSV in this folder: https://sleepdata.org/datasets/cfs/files/datasets
This is the variable/column you'll want: https://sleepdata.org/datasets/cfs/variables/stloutp
We plan to update our annotation files in the future to include these timing variables. This is a limitation of the Compumedics Profusion XML exports.
Thanks for asking!
Thanks for raising this issue. Please check out the Hypopnea event tags section on this page: https://sleepdata.org/datasets/mros/pages/polysomnography-introduction.md
In short, look for events tagged as "Unsure" in the XML files. These are also hypopneas that can be included in the numerator for AHI metrics. There appear to be 353 Unsure events for aa0181 at Visit 2.
How are you extracting the actigraphy (CSV rows) and PSG (XML SleepStaging) data points you need?
Here's one way to think about the task of pulling the data that overlap. Let's say I found a 10 hour PSG recording that contained 1200 epochs of staging data (30 seconds in length). For that subject I could then lookup the overlap line number from the epoch-by-epoch CSV file and select the next 1200 actigraphy epochs to set beside the PSG epochs. Perhaps you are choosing extra actigraphy epochs somehow? Maybe it also has to do with the slight time differences (e.g., actigraphy line times always end in :00 or :30, whereas PSG would not normally start at those times)?
If I had to guess, the SSC (Stanford Sleep Cohort) subjects probably have some of the "fullest" PSG montages. The MNC montage info page is a collection of all the signals found across the cohorts.
The folders correspond to different cohorts that contributed to the MNC project. You can find additional (limited) information in the primary MNC publication: https://pubmed.ncbi.nlm.nih.gov/30523329/
The MNC team has not shared scoring annotation files for most of the cohorts.
Your access was granted to the mesa-commercial-use dataset (e.g., nsrr download mesa-commercial-use), so make sure your NSRR gem commands are formatted for that version of the MESA data.
Thanks for using the site. I see you have access to the Sleep Heart Health Study (SHHS).
You can read about the study protocol in these links:
Subjects were chosen from existing observational studies. There were two main visits/waves in SHHS, so some subjects have two overnight sleep studies collected. The first wave was between 1995-1998 and the second wave between 2001-2003.
Thanks for using the site. I believe the STAGES device, the Amazfit Arc (https://amazfit.bg/arc.html), did not output "raw" accelerometry data, but rather data binned at 30-second or 60-second epochs (I can't recall; haven't looked at data recently). Presumably the "activeness" column is some metric of movement, i.e., higher levels mean more activity. The device (evidently) did not output estimations of sleep vs. wake epochs.
Perhaps the MESA and HCHS dataset would be more appropriate for you. These datasets do not include raw accelerometry data either, though the 30-second epoch-by-epoch data do contain sleep-wake estimates. MESA includes concurrent PSG for many subjects.
We hope to share truly "raw" accelerometry data in future cohorts.
The HomePAP dataset split night studies are the only overnight recordings with CPAP that come to mind. All the sleep scorers in HomePAP were trained polysomnologists from American Academy of Sleep Medicine (AASM) accredited sleep centers.
Thanks for using the site!