Sleep log and Not worn

[Callum Regan]

Hi all, 

I have recently processed gt3x data (n = 60) from hip-worn actigraphs that weren't worn during the night. I have successfully used an advanced sleep log 

However, I see consistently that the guider used for sleep onset on the last day is always "NotWorn+invalid" rather than the sleep log. This also occasionally happens for the first wake. I fear that I could lose out on valuable data. 

Attached is an example of one participant with the file summary report (I wasn't able to attach a meta file or script at the same time). For instance for the 1st day (Thursday) the sleep log says that wakeup is at 08:30 but the guider "NotWorn" uses 14:00. And for the last day (Wednesday) the sleep log states that sleep onset is at 21:00 but the "NotWorn" guider uses 17:30.

My script commands for part 4 are as follows: 

includenightcrit = 0, 
     def.noc.sleep = c(1),
     outliers.only = FALSE, 
     criterror = 3, 
     part6CR = TRUE,
     part6HCA = FALSE,  
     loglocation = ""
     relyonguider = TRUE, 

     do.visual = TRUE,

HASPT.algo = c("NotWorn"), 

 HASIB.algo = "NotWorn",
 HASPT.ignore.invalid = NA,

If anyone has any advice I would very much appreciate it.

Many thanks,

Regards,

Callum

[Callum Regan]

Here is the meta file and sleep log for this participant. I also attached the full script. 

Also, I tried changing  HASIB.algo to "VanHees2015" and the same output was produced. 

Many thanks, Callum.

[Vincent van Hees]

Hi Callum,

Thanks for sharing your GGIR code and example files.

Your diary ends at sleep onset for the evening of 30/10, but does not include wake-up time for 31/10. GGIR needs both in order to use the diary. If this is a problem, you could impute in your sleep diary the missing timestamp.

By the way, the visualisation you shared indicates that the accelerometer is never worn during the night and also hardly worn during daytime. To me this sounds like a good justification to skip this recording from further analysis.

Best, Vincent

--
You received this message because you are subscribed to the Google Groups "R package GGIR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to RpackageGGIR...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/RpackageGGIR/e1d51265-840a-46a8-b8ec-62a90cfe8456n%40googlegroups.com.

[Callum Regan]

Hi Vincent,

Thank you kindly for your response and the explanations provided.

I understand now regarding the sleep diary. For this person the 30/10 was the last accelerometer wear day and thus they didn't wear it on the 31/10 which is why there was no wake-up time for the 31/10.

With your advice, I imputed wake-up/sleep onset times the day before the first wear day and the day after the last wear day and the sleep log was used as the guider. When I did this one less day was valid in part 5 which looks to be an accurate classification. 

I noticed that for most participants in my sample when NotWorn was used as the guider for the first wake-up and last sleep onset this matched that of the sleep log but perhaps it could be wise to impute values in order to use the sleep log for all days and for example like this were there was a mismatch. 

Thank you for highlighting this. 

With regard to the visualisation, participants were not instructed to wear the accelerometer during the night so this non-wear time can be expected.

I'm assuming that the dark blue (no movement sib daytime) indicates non-wear during the day? (I've struggled to find information for the outputs for the new file summary report)
And yes the acceleration signals indicate that there was not much movement during these days and the Part 5 results suggests that large proportions of these days were spent inactive.

I wonder, given that output for 5 days in part 5 was produced, with dur_day_min all over 600 mins, using the 2/3 waking day criteria, could these days be considered as very inactive rather than not-worn? 

However, this participant did have one of the lowest ACC_day_mg_pre = 8 - averaged over the days, so perhaps there should be a cut off for outliers that have extremely low movement...

Any further advice on this would be most appreciated and thanks again for the advice thus far.

(Attached is the new file summary report for clarity)

[Vincent van Hees]

Hi Callum,

Your email is long and not clearly structured. Keep in mind that software does not mean that my time is also for free. I only provide brief responses to clearly formulated questions for free.

If you would like me to spend more time on your questions then I would like to propose that you or your employer hires me as consultant.

Thanks,

Vincent

Dr. Vincent van Hees | Independent consultant | https://accelting.com/

To view this discussion, visit https://groups.google.com/d/msgid/RpackageGGIR/965d6620-6485-457d-969e-2a7dd8bb4ca6n%40googlegroups.com.

[Callum Regan]

Hi Vincent,

Point taken. Thank you for your support and guidance. 

All the best, 

Callum