Question regarding do.imp effects on inactivity and variable selection for average wear time in GGIR

[Malontax]

When I run the package with do.imp=FALSE, I notice that participants have a higher dur_day_total_IN_min and lower values for other intensities, such as dur_day_total_LIG_min and dur_day_total_MOD_min. Am I correct in assuming that when do.imp = FALSE, non-wear time is classified as inactivity? Does setting do.imp=FALSE therefore possibly cause your total inactivity to be higher?

Additionally, I was wondering how to best report participant compliance or average wear time. I am finding it difficult to select the right variables. Since the non-wear percentage in Part 5 only includes data that meets the '>16 hours valid data' criterion, should I refer to the 'N valid hours' from Part 2 instead?

I hope these questions are clear, and I look forward to your insights.

[vincentvanhees]

Correct, not imputing non-wear time means that the lack of movement typical for non-wear time is treated as inactivity.

N valid hours from GGIR part 2 output tells you about protocol compliance by study participant, in its most basic form: Is the sensor worn during the time when we expect the sensor to be worn. I think this is useful when compliance is your main outcome of interest.

However, if you only want to report how many valid days/nights you have in your main analysis that primarily depends on GGIR part 5 outcomes then I would report data availability specific to part 5.

[Malontax]

Thank you for your response!

To get an idea of how dedicated the study participant is to wearing the accelerometer, I need to know how compliant they are. Is the 'wear_dur_def_proto_day' from Part 2 useful for determining for the compliance, or should I focus on N valid hours?

Finally, I noticed that the measurement does not go up to 6.99 for all participants, even though it should. Could this be because the accelerometer stopped measuring prematurely, for example due to a dead battery or full memory?

[vincentvanhees]

You would need to look at two output variables:

meas_dur_def_proto_day represents the observed measurement duration defined by the study protocol. In other words, if the study protocol is to wear the sensor for 7 days and if you collect 9 days of data then meas_dur_def_proto_day = 7, but it will be 4 if only 4 days of data are collected.

wear_dur_def_proto_day the amount of time spent wearing the device during meas_dur_def_proto_day.
wear_dur_def_proto_day tells you about compliance within the context of the study protocol, but is only meaningful in the context of meas_dur_def_proto_day.

So, I would present the difference or the ratio between both as actual indicator of participant compliance to the study protocol.

[vincentvanhees]

If you have suggestions on how the GGIR documentation can be updated to better explain this then please help.

[vincentvanhees]

@Malontax can this be closed or do you have suggestions for improving the GGIR documentation?

[Malontax]

Dear Vincent, Sorry for my late reply! For now I do not have a direct suggestion for the improvement of the GGIR documentation. Kind regards, Malon Tax

…

________________________________ Van: Vincent van Hees ***@***.***> Verzonden: donderdag 16 april 2026 16:41 Aan: wadpac/GGIR ***@***.***> CC: Malon Tax ***@***.***>; Mention ***@***.***> Onderwerp: Re: [wadpac/GGIR] Question regarding do.imp effects on inactivity and variable selection for average wear time in GGIR (Discussion #1470) @Malontax<https://github.com/Malontax> can this be closed or do you have suggestions for improving the GGIR documentation? — Reply to this email directly, view it on GitHub<#1470 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A52VDVFQKO5QF6HFCT26CHT4WDWLHAVCNFSM6AAAAACV7M3NHSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMNJYHA3TANI>. You are receiving this because you were mentioned.Message ID: ***@***.***>