We've updated our privacy policy.

New approach diagnosing Apnea

2 posts
Was this reply useful? Learn more...
turtility +0 points · 13 days ago

Dear All,

I am researching for an issue on apnea cases.

My question is: Is that possible to diagnose apnea basically from the SaO2- SpO2 values only ? By this, we will figure out if there are reasonable results and make people more comfortable to get apnea treatment from their houses, especially whom are disabled.

I use the polysomnography .edf raw values, filter Time-SaO2 with EDF browser and import to MS SQL.

In MS SQL I run a query code and calculate the oxygen desaturation during all time which is appoximately 8,5 hours for each, around 30.000 secs.

I focus on the AHI and RDI basically. As I know shortly, RDI = AHI + Arousals .

When I compare the real study results ( lets consider shhs1-200041_filtered_data ), non-apnea results generally match but existing-apnea results and my own results do not seem mathcing.

I wonder, is the approach has something to add, modify or change ? Can I get some reasonable and undestandable results with this methodology.

Beside, I can share the sql query code and results if anyone wants ?

I really appreciate to hear from experts as you.



2 posts
Was this reply useful? Learn more...
turtility +0 points · 13 days ago

In addition, image and text does not match with the nssrid and edf files. Please refer to image. you can considershhs1-200001_filtered_data, shhs1-200002_filtered_data and shhs1-200003_filtered_data.

Thanks in advance

160 posts
Was this reply useful? Learn more...
mrueschman +0 points · 6 days ago


The RDI* variables in SHHS are not necessarily "AHI + arousals". They are more like AHI variables, but all events (apneas and hypopneas) are required to be associated with a desaturation (of 4%, for instance, in RDI4P).

I assume there are people trying to "diagnose" sleep apnea using only an oximetry channel. Naturally, the ODI closely aligns with AHI in many cases.

Do you have any specific data questions? Or inconsistent data findings? I wasn't sure exactly what I was looking at in your image.

Note that if you derived "new" oxygen desaturations from the raw signal then you would want to be sure your methodology matched with whatever was used by the original sleep scoring software. Otherwise your new metrics would not be comparable.