Related
Options

Epix 2 vs Sleep Study: SpO2 off, wake & deep undercount

Etupes25
over 3 years ago

I just went through a medical sleep study using a WatchPAT like device and I was curious to see how the Epix 2 metrics would compare.

Overall, with the Epix 2

- sleep time is overestimated, as wake time is only detected half of the time

- light sleep is in the ball park

- deep sleep is underestimated, as it is detected half of the time, and replaced with mostly light sleep, sometimes REM sleep

- Sp02 was mostly off (either way too high or way too low) for the first 2/3 of the night. It was more accurate during the last 1/3. Unfortunately, the highest and lowest points didn't match for the most part.

- Heart rate was OK, but lowest points were not recorded

Epix2 (v8.16) Sleep Study
Total duration 8:19:00 8:18:00
  Sleep time 7:53:00 7:28:00
  Wake time (mn) 26 49
  # wake periods 2 7
  wake time / duration 5.21% 9.93%
Sleep phases
  REM  / Sleep 31.08% 17.25%
  Light Sleep / Sleep 62.37% 69.05%
  Deep Sleep / Sleep 6.55% 13.70%
  Total 100% 100%
SpO2
  Lowest SpO2 86% 83%
  Maximum SpO2 100% 100%
  Average SpO2 95% 96%
Heart Beat
  Lowest 47 37
  Maximum 74 79
  Average Heart (bpm) 52 52

  • 0 over 3 years ago in reply to trux

    Thank you, this is a great find. The paper you linked states that sleep stage classification accuracy of the Firstbeat ECG- based algorithm is between 63% and 66%. It means that 2 times out of 3, when the Firstbeat ECG HRV- and movement-based algorithm classifies a 5 second period of sleep, the result concurs with the experts' polysomnography classification (who apparently have a ~90% agreement rate themselves). 1/3 mismatch rate is "reasonably good", perhaps, as the paper states. Garmin watches are not ECG devices, so we should expect a lower match.

    In my personal case, I get consistently about 50% of my deep sleep when I am awake watching TV sitting in the living room,  before going to bed.

    On to looking for SpO2 measurements validation...