Data and code used for 'Thigh-Worn Accelerometry: A Comparative Study of Two No-Code Classification Methods for Identifying Physical Activity Types'

This repository contains all data necessary to reproduce the results for the manuscript titled 'Thigh-Worn Accelerometry: A Comparative Study of Two No-Code Classification Methods for Identifying Physical Activity Types'.

File structure

- analysis

The analysis subfolder contains all R scripts used for the study:

1. Sample size estimation
2. Synchronisation of the timestamps
3. Processing of the raw data
4. Calculating the interrater agreement
5. Calculating all performance metrics and producing the plots

In addition, the two subfolders contain the plots and result tables produced when running the scripts.

- data

The data folder contains all raw data as well as the processed data. A subfolder for each subject contains the video annotations (.eaf), the raw acceleration data (.csv) and the SENS motion classification data (.csv).

The ActiPASS subfolder contains the raw acceleration files in binary file format as well as the ActiPASS output.

The shiny subfolder contains the R code used for running the shiny app during data collection as well as the logged data and timestamps.

- documents

This folder contains any additional documents used in the study.

Requirements

The data processing and analysis was performed in R (Version 4.3.2). To run the full analysis in R, the following packages need to be installed:

• tidyverse
• xml2
• lubridate
• here
• dygraphs
• hms
• irr
• yardstick
• cowplot
• gt