I'm wondering if anyone has developed a script to run in the R statistical package that can validate data files (i.e. sufficient wear time), clean files (remove non-wear time etc.) and produce the usual summary estimates (i.e. cpm, mins of mvpa etc.)?
We have published an algorithm for wearing/nowearing. It is described in
Validation of accelerometer wear and nonwear time classification algorithm. Choi L, Liu Z, Matthews CE, Buchowski MS.
Med Sci Sports Exerc. 2011 Feb;43(2):357-64
The algorithm is freely available as an R package “Physical Activity”, downloadable from http://www.r-project.org/.
We obviously love it and it performs much better than ActiLife. Please let me know if you would have any problems with the application.
With regard to Actigraph raw data: I developed R-package GGIR for multi-day raw data collected with Genea and GENEActiv, http://cran.r-project.org/web/packages/GGIR/index.html
I could expand this package with Actigraph raw data functionality if people are interested....
The "PhysicalActivity" package developed by Choi and colleagues (including Mac - above) is quite helpful and has a lot of functionality. That said, there are some nuances to properly and efficiently processing such data using this package - depending upon the exact structure of your data (e.g., what file types you trying to process [.agd, .dat, .csv]), and from what generation ActiGraph the data was collected (e.g., GT1M, GT3X, GT3X+, etc.).
I have developed some custom R coding that may be appropriate for your needs. Shoot me a message if you are interested.
Thanks Mac, Vincent and John. I'm new to coding/syntax so am very much a copy and paste and then modify the essential parts user...
Mac's package seems straightforward enough to me and also gives lots of functionality. However I would be a bit stuck with what to do next to produce summary statistics...and also how to deal with the generations of AG we have in our project data (i.e. GT1M, 3x and 3x+).
I have MeterPlus but this is too time consuming to use! Actilife seems okay but is missing some key functions (i.e. being able to differentiate between weekday and weekend data) when determining wear time etc...
Thanks Vincent but I'm working with count level data as there aren't any raw data cutpoints that I know of in kids?! and I don't understand the signal processing stuff :)
Cheers
Ash
- Badges
- Science topic
- Similar topics
- Medicine
- Epidemiology