---
title: "Research accelerometers: ActiGraph and Axivity"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Research accelerometers: ActiGraph and Axivity}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>", eval = FALSE)
```

```{r setup, eval = TRUE}
library(PhysioDevices)
```

`readGT3X()` and `readCWA()` read the two field-standard raw research
accelerometers -- **ActiGraph** (`.gt3x`, common in US cohorts) and **Axivity
AX3/AX6** (`.cwa`, used by UK Biobank and other large studies) -- into the same
tri-axial-acceleration `PhysioExperiment` that `PhysioWearable::readAccelCSV()`
produces, so they flow straight into the free-living pipeline.

The binary decode is delegated to the validated `read.gt3x` and `GGIRread`
packages (install them first); the readers only adapt the result.

```{r read}
pe <- readGT3X("subject.gt3x")          # or readCWA("subject.cwa")
pe                                       # a PhysioExperiment: acceleration (x/y/z, g)
samplingRate(pe)
```

## Free-living metrics

The acceleration assay feeds ENMO, non-wear/calibration, intensity and the
free-living summary exactly as for `readAccelCSV()` output:

```{r freeliving}
acc  <- SummarizedExperiment::assay(pe, "acceleration")
enmo <- PhysioWearable::computeENMO(acc, sampling_rate = samplingRate(pe))
PhysioWearable::summarizeFreeLiving(enmo)          # wear time, MVPA, intensity gradient, ...
```

## Notes

- `read.gt3x` / `GGIRread` are `Suggests` -- each reader errors with an install
  hint if its package is missing.
- `readCWA()` reads the whole file by default; pass `start`/`end` (block range)
  to read a long recording in pieces.
- These devices record **raw acceleration only** (no HR/SpO2); pair them with a
  smartwatch reader for those signals.