shiny-meta-models

Shiny dashboard to validate and explore probabilistic inputs and outputs of health economic models, using metamodelling methods

demo

A live demo of the PACBOARD application can be found at https://bdsi.shinyapps.io/pacboard/

development

In a terminal, clone the repository and cd into it:

• git clone https://github.com/BDSI-Utwente/shiny-meta-models.git
• cd shiny-meta-models

Initialize the pacheck submodule:

• git submodule update --init --recursive

Open the project in your favourite editor (probably RStudio), then install the required R packages:

• open dashboard/dashboard.Rproj
• in an R console, run renv::restore()

Note that at time of writing, the MRAN repository used by renv for historical binary versions of packages is in limbo, and renv::restore() may fail. If this occurs, setting options(renv.config.mran.enabled = FALSE) may solve the problem, but will take much longer to complete as historical binaries will have to be compiled from source.

You can now start a local server directly from the RStudio UI;

• open dashboard/app.R, then click the Run app button

or from the R console by running:

• shiny::runapp("app.R")

The local server should automatically update and refresh on changes, but due to a bug/quirk/feature in how shiny handles file changes, only changes to the main app.R file are registered. Setting up an external file watcher to trigger updates may be beneficial for an extended development workflow, but is left as an exercise for the reader.

Project Structure

The dashboard/ folder contains the source code for a shiny frontend to the pacheck package (which is referenced as a submodule in the pacheck/ folder).

The explore/ folder contains scripts and snippets used while developing and testing.

Application logic is contained in dashboard/app.R, with handler functions for the specific tabs located in dashboard/R/, and various helper functions in dashboard/functions/. Most of the core functionality underlying the dashboard is provided by the pacheck package.