NEWS

---

rester 0.1.0

• Initial CRAN submission.
• Initial version wrapping the AcceleRest model (https://github.com/NielsRLorenzen/AcceleRest), with a citation to Lorenzen et al. (2026) doi:10.64898/2026.01.28.26345056.
• Added ar_download_repo()/ar_extract_repo() and ar_download_weights() to download and extract the model code and pretrained weights, and ar_setup() to do both.
• Added estimate_rest() to predict sleep stages and respiratory events.
• .onLoad() declares the Python requirements with reticulate::py_require().
• Device files are read with actiread::acti_read(), and data are standardized, given a sample rate, and gap-filled with actibase. Compressed files the reader cannot handle are decompressed to a temporary file, which is deleted after reading.
• On macOS, KMP_DUPLICATE_LIB_OK=TRUE is set (if unset) so torch and OpenMP-using readers (e.g. GGIRread) can run in the same R session.