Code base for deriving movement sequences from accelerometer (Actigraph) data and mapping them to symbolic classes, as originally described in: Chin A Paw et al. MSSE 2019 (https://doi.org/10.1249/mss.0000000000001849). The original version of the code was written by Xinhui Wang. We are currently improving this repository to become more flexible to alternative ways of pre-processing the data.

What does this repository do?

• File main_script.R shows how the code is used.
• Store the Actigraph count data .csv files in one directory, and provide the required user input.
• When you run the script, the function run_pipeline ties together the following functions to generate sequence maps:
  1. a) The accelerometer data is loaded (using function read_file); b) The accelerometer counts are aggregated (using function resample);
  2. Sequence maps are generated (using function generate_sequence): a) After the non-wear time (using function 'detect_nonwear') and invalid days are filtered out; b) For each epoch in a data file the cut-point intensity classes, followed by the lengths of the consecutive epochs in the same cut-point class (bout lengths) and their corresponding values are determined (using function detect_bouts); c) Then the detected segments (bout values and lengths) are labeled with the corresponding symbols (using function add_symbols); d) Finally, the recording level (using function 'structure_per_recording') and day level (using the function 'structure_per_day') sequence maps are generated.

Note for developers: Run script run_roxygen2.R to update the documentation files in the man folder.