Average acceleration and intensity gradient with missing data

[Natã Gomes De Lima Stavinski]

Hi Dr. Van Hees and GGIR team,

I hope this message finds you well.

I am currently working with a survey using GGIR to analyze mainly PA (although my code presents the sleep output also) and I would like to use the average acceleration and the intensity gradient as my outcomes of volume and intensity of PA in different parts of the day (morning and evening). However, I believe that some accelerometers (or many) do not have complete data (24h) in most days (some might not have a single day with 24h of recording). Yet, most of the missing data is on the “sleeping part” of the 24h, which I do not intend to use. Am I able to produce reliable outputs with this kind of data? If yes, how can I do so?

Lastly,  some of these data is only in .agd format. Am I able to use it or only the .gt3x format is usable?

 

Cheers,

Natã

 

Ps: my code now is the following:

#Loading the packages

 

library (GGIR) #Package to read accelerometry data

library(read.gt3x) #Read ActiGraph gt3x files

library(ActCR) #Only necessary if you're using circadian rhytm indicators data

 

GGIR(

  datadir = "H:/Meu Drive/Doutorado/Data",

outputdir = "H:/Meu Drive/Doutorado/Results",

 

 

#==================

# Part 2

#==================

 

strategy = 1,

hrs.del.start = 0,          hrs.del.end = 0,

maxdur = 7, #max number of days expected in a data file based on the study protocol

includedaycrit = 16,

qwindow=c(7,11,12,19), # I believe this is what defines the time I wanna use

mvpathreshold =c(100),

excludefirstlast = FALSE,

includenightcrit = 16,

iglevels = TRUE,

 

#==================

# Part 3 + 4

#==================

 

def.noc.sleep = 1,

outliers.only = TRUE,

criterror = 4,

do.visual = TRUE,

 

#=====================

# Part 5

#=====================

 

threshold.lig = c(30), threshold.mod = c(100),  threshold.vig = c(400),

boutcriter = 0.8,      boutcriter.in = 0.9,     boutcriter.lig = 0.8,

boutcriter.mvpa = 0.8, boutdur.in = c(1,10,30), boutdur.lig = c(1,10),

boutdur.mvpa = c(1),

includedaycrit.part5 = 2/3,

 

#=====================

# Visual report

#=====================

 

timewindow = c("WW"),

visualreport=TRUE)

 

[Jairo Hidalgo Migueles]

Hi Natã,

Regarding the output, you can get the output in part 2 of GGIR even if participants did not wear the accelerometer during sleep. And even parts 3, 4 and 5 can work with that data. If the output is reliable or not would be something not directly related to GGIR but to your protocol, consider participants forget wearing the accelerometers after wake up, then you would be missing a part of their day which might influence the average acceleration. This is something that affects all the protocols and outputs from studies based on collecting data only during awake time.

As for you other question, GGIR is now able to read some epoch aggregates from actiwatch and ukbiobank, but actigraph agd format is not one of file format that GGIR can read at the moment.

Best,
Jairo