healthcareai

The aim of healthcareai is to make it easy to do machine learning with healthcare data. The package has two main goals:

• Allow one to easily develop and compare models based on tabular data, and deploy a best model that pushes predictions to either databases or flat files.

• Provide tools related to data cleaning, manipulation, and imputation.

Install the latest release on Windows

Open RStudio and work in the console

install.packages('healthcareai')
install.packages('assertthat')

How to install the latest version on macOS

Note: If using macOS with healthcare.ai, you'll have to use csv files. We're working on adding MySQL connections. We'd love to hear which other databases your connecting to, so we can provide native support!

• Open the Mac Terminal
• Install Xcode compilers via xcode-select –install
• Accept the Xcode license via sudo xcodebuild -license
• Install Homebrew (the macOS package manager) with
• Install ODBC driver via brew update && brew install unixODBC
• Open R Studio
• In the console, install RODBC from source with install.packages('RODBC',type = "source")
• In the console, install other R healthcare.ai prerequisites via

install.packages(c('caret','data.table','doParallel','e1071','grpreg','lme4','lubridate','pROC','R6','ranger','ROCR'),repos = "https://cran.cnr.berkeley.edu/")

• Install healthcare.ai

install.packages('healthcareai')
install.packages('assertthat')

How to install latest version on Ubuntu (Linux)

• Follow steps 1 and 2 here to install R
• Run sudo apt-get install libiodbc2-dev
• Run sudo apt-get install r-cran-rodbc
• After typing R run install.packages('healthcareai') and install.packages('assertthat')

Install the bleeding edge version (for folks providing contributions)

• Grab prerequisites via the console of RGui or (preferably) RStudio

install.packages(c('caret','data.table','devtools','doParallel','e1071','grpreg','lme4','lubridate','pROC','R6','ranger','ROCR','RODBC'),repos = "https://cran.cnr.berkeley.edu/")

library(devtools)
devtools::install_github(repo='HealthCatalystSLC/healthcareai-r')

Tips on getting started

Built-in examples

Load the package you just installed and read the built-in docs

library(healthcareai)
?healthcareai

Website examples

See our docs website

Jupyter notebook examples

If you like Jupyter notebooks, check out step 1 and step 2 in model building with healthcareai.

Join the community

Read the blog and join the slack channel at healthcare.ai

What's new?

Since the CRAN 0.1.11 release, the following has been added to the bleeding edge version

• Output nightly predictions to a dataframe for use with MySQL, Oracle, etc. See the csv example at ?RandomForestDeployment
• Evaluate in-the-wild performance via AU_ROC and AU_PR scores and plots. See more at ?generateAUC

For issues

• Double check that the code follows the examples in the built-in docs

library(healthcareai)
?healthcareai

• Make sure you've thoroughly read the descriptions found here

• If you're still seeing an error, file an issue on Stack Overflow using the healthcare-ai tag. Please provide

  • Details on your environment (OS, database type, R vs Py)
  • Goals (ie, what are you trying to accomplish)
  • Crystal clear steps for reproducing the error

Contributing

You want to help? Woohoo! We welcome that and are willing to help newbies get started.

First, see here for instructions on setting up your development environment

After that's done, here's the contribution workflow:

1. Identify an issue that suits your skill level here

   • Only look for issues in the Backlog category
   • If you're new to open source, please look for issues with the bug low, help wanted, or docs tags
   • Please reach out with questions on details and where to start

2. Create a topic branch to work in, as described here

3. Create a throwaway file on the Desktop (or somewhere outside the repo), based on an example

4. As you make changes

   • Document any new functions, methods, etc via roxygen2
   • Be sure to build often
     • Note: building checks for errors and allows you to use the latest code
   • Make small commits after getting a small piece working
   • Push often so your changes are backed up. See here for more

5. Early on, create a pull request such that Levi and team can discuss the changes that you're making. Conversation is good.

6. When you're done with the issue you chose, do the following

   1. Merge the master branch into your topic branch (so that you have the latest changes from master)

   git checkout LeviBugFix
   git fetch
   git merge --no-ff origin/master
   

   • This opens a text editor called vim, where you type i, type your commit message, and then save

   2. Build and fix any errors

   3. Run tests via devtools::tests() or CTRL+SHIFT+D and fix any errors

   4. Run the roxygen2 examples via devtools::run_examples() and fix any errors

   5. Under the build tab, run 'Check' and verify that only one roxygen warning arises

      • This warning is due to the limitations of roxygen and R6 method documentation
      • This is the only warning/error/note that's allowed when merging to master

   6. Now that your changes are working, communicate that to Levi in the pull request, such that he knows to do the code review associated with the PR. Please don't do tons of work and then start a PR. Early is good.