mrueschman

Hi turtility,

You could pull an AHI variable (using your preferred definition) from the SHHS dataset and use this as a stand-in for sleep apnea diagnostic criteria (e.g. 5-15 mild, >15 moderate, >30 severe). Here's a search to some SHHS AHI variables: https://sleepdata.org/datasets/shhs/variables?search=ahi_

These were research sleep studies, so we don't have reports/diagnoses, per se.

Mike

Hi,

Thanks for checking out the NSRR. Just curious, is it still slow if you try to manually download a file through your web browser, e.g. from here https://sleepdata.org/datasets/shhs/files/polysomnography/edfs/shhs1 ?

Hi - thanks for checking out the site. I think it will be possible to generate a supplemental dataset to meet your needs. I responded to the email you sent to support@sleepdata.org and I think we can sort out the details there.

No, there isn't a variable like that. These were all research studies, so neither the patient nor their sleep study were judged/assessed by a health professional.

Yuval,

Unfortunately I don't have any grand explanations to report. My colleague looked at the source data from some of your examples and the supine TST appeared to match your findings (i.e. differed from the CSV dataset as well). I would go with your findings in this case since that is what the raw signal is telling you.

It's hard to explain things in SHHS because the data were exported from versions of the scoring software more than 20 years ago. That software no longer works and we have backwards compatibility issues when we try to open these data from these 1990s in modern software to try and sort out how these issues occurred. Clearly some "bad" data have made their way into the CSV/results dataset.

Sorry I couldn't be of more help. Thank you for being diligent in your review of the data and good luck with the analysis!

Mike

By the way, could you quantify a "few mismatches" across the 5,804 SHHS1 studies? You said they were sometimes identical, but I'm guessing there are many that are very close to one another but not quite exact. I have seen that sort of pattern (close but not exact) when doing similar explorations of the data in the past.

Maybe you could derive how many subjects have differences of > 5 minutes?

Shanshan,

Thanks for checking out the resource. I emailed another NSRR team member to get her feedback. My inkling is that this sort of pattern is within the normal range. I believe people can have many short awakenings over the course of a night and not "remember being awake", so to speak, in the morning.

Hmm, I still don't fully understand. These were research subjects, so there won't be any sort of subjective "interpretation" (by a medical professional).

The polysomnologist who scored the study gave ratings of the quality of individual signals and the overall quality of the sleep study, e.g. https://sleepdata.org/datasets/shhs/variables?folder=Measurements%2FPolysomnography%2FSignal+Quality

Sleep efficiency (https://sleepdata.org/datasets/shhs/variables/slp_eff) is a commonly used metric to assess the "quality" of a person's sleep. This is the proportion of time spent asleep within the time in bed. Higher efficiency is thought to often indicate a "better" night's sleep.

Hi Yuval,

I'll take a look and get back to you sometime next week.

Mike

Were you thinking of self-reported sleep quality? Maybe these variables from the Morning Survey in SHHS2 would help: https://sleepdata.org/datasets/shhs/variables?folder=Questionnaires%2FSHHS2%2FMorning+Survey

If you had another sort of quality in mind let me know and I can comment again.