/ccdata

Essential tools and utility functions to facilitate the data processing pipeline, data cleaning and data analysing of clinical data from CC-HIC

Primary LanguageRGNU General Public License v3.0GPL-3.0

Build Status codecov AUR

The ccdata R package is the centralised tool set for the critical care data analysis. Three key components can be found in the current version of ccdata package,

  • The XML parser
  • Data cleaning and validation modules
  • Table exporter
  • Data analysis functions

The ccdata package is portable to all platforms where a R environment is available. It covers the most part of the data processing pipeline. The XML files will be parsed to an R data structure which is a bespoken query-able storage format for the critical patient records. With the selecting and cleaning process, the user can obtain a clean table guided by the YAML configuration file specified by the users. The user can subsequently perform their data analysis on the clean table.

Data processing pipeline

The YAML configuration example:

NIHR_HIC_ICU_0108:
  shortName: hrate
  dataItem: Heart rate
  distribution: normal
  decimal_places: 0

  # filter1: do not use the episode where hrate cannot be found.
  nodata:
     apply: drop_episode

  # filter2: mark all the values based on reference range (traffic colour)
  # remove entries where the range check is not fullfilled.  
  range: 
      labels:
          red: (0, 300)
          amber: (0, 170) 
          green: (50, 150)
      apply: drop_entry

  # filter3: compute the item missing rate on given cadences; in this case, we compute the daily (red) and hourly (amber) missing rate, and only accpet episodes of which hourly missing rate (amber) is lower than 30%. 
  missingness: 
      labels:
          red: 24
          amber: 1
      accept_2d:
          amber: 70 
  apply: drop_episode

Required packages

  • R (>= 2.1.0),
  • XML,
  • reshape2,
  • data.table,
  • yaml,
  • pander,
  • RPostgreSQL,
  • sqldf,
  • testthat,
  • data.tree

How to install the R package

Mac & Linux

git clone git@github.com:UCL-HIC/ccdata.git
R CMD INSTALL ccdata # "sudo R CMD INSTALL ccdata" if root access is required.

RStudio

  • Download the tar file from ccdata Github page.
  • In the package panel click the button install.
  • select Package Archive File (.tgz, .tar.gz) for the select from tab.
  • click install

How to contribute

The ccdata package is currently underdevelopment. We wellcome users using, commenting about the code on the master branch. If you have any question, you can just raise an isssue on Github or contact the developers via email (s.shi@ucl.ac.uk). Please let us know if you also want make contribution to the code development.