The GGIR 3.3-6 release is now on CRAN. In this blog post I will talk you through the main updates since the 3.3-0 CRAN release from September 2025. For a full overview see the GGIR changelog, where # numbers refer to specific work items in the GGIR GitHub repository.
General updates
- The GENEActiv accelerometer allows for collecting data at 85.7 Hertz. This configuration has been successfully used for many years. However, in the GGIR code we have made assumptions that the number of samples per epoch is a round number and not a fractional number, resulting in some ambiguity in the calculation. To address this ambiguity in calculation, GGIR now automatically first resamples the 85.7 Hertz signal to 86 Hertz before processing it. For this, GGIR uses the same resampling function as used for Axivity data (GGIRread::resample to be specific).
- QC plots generated in GGIR part 2 used to be stored in pdf files of up to 300 pages. This has now been replaced by a single png file per recording. I believe this makes the visualisations easier to work with and at the same time has allowed for speeding up the report generation.
New functionality
- Step/cadence functionality has been expanded with a variety of output variables as now documented in a new vignette. Credits for funding this enhancement: Ben Maylor and Alex Rowlands.
Bug fixed
- Bug #1415 triggered an error message when the main SPT window was directly preceded or followed by candidate sleep windows.
- Bug #1404 reflects an incorrect calculation of Oakley1994, Galland2012, and ColeKripke1997. By mistake epochs before and after the epoch of interest were weighted in reversed order.
- Bug #1400 triggered an error when running GGIR part 6 with parameter ‘do.parallel = TRUE’.