RtEval

Outline

The goal of this repository is to evaluate software packages for estimating the effective reproduction number.
We do this by fitting models from a range of software packages to simulated and real data
In the future, these data will be stored in another repository (“rtdata”)
In the future, we will summarise the output of each package consistently using the summrt R package
We hope that this repository will contain evaluations which are run semi-regularly using GitHub actions so as to serve as a living, rather than static, comparison

Live vignettes

The following vignettes (which are hosted under the vignettes folder) were automatically built using GitHubActions:

epiestim: https://epistorm.github.io/RtEval/epiestim_vignette.html
EpiLPS: https://epistorm.github.io/RtEval/EpiLPS_vignette.html
EpiNow2: https://epistorm.github.io/RtEval/EpiNow2_vignette.html
RtEstim: https://epistorm.github.io/RtEval/RtEstim_vignette.html
Joint: https://epistorm.github.io/RtEval/eval_vignette.html

Structure

Remains to scope!

Set-up

To run the code in this repository you’ll need at least R version 4.0.0:

required_version <- "4.0.0"
if (getRversion() < required_version) {
  stop(paste("Your R version is", getRversion(), ". Please update to version", required_version, "or higher."))
} else {
  cat(paste("R version is sufficient:", getRversion(), "\n"))
}

You will also need these packages:

install.packages(
  c(
    "rmarkdown", "shiny", "knitr", "plotly",
    "purrr", "splines", "tidyverse", "readr",
    "EpiLPS", "EpiNow2", "lubridate", "rstan",
    "cmdstanr", "R0", "remotes", "ggpubr",
    "ggtext", "lemon", "patchwork"
  )
)

remotes::install_github("dajmcdon/rtestim")

install.packages("EpiEstim", repos = c("https://mrc-ide.r-universe.dev", "https://cloud.r-project.org"))

Rt Collabathon 2024

This work is a part of the Epistorm Rt Collabathon.

Participants in the workshop may be interested in these links:

ToDo:

Some remaining todo’s include:

having some distributions vary over time in the simulated data (e.g., mean reporting delay gets shorter over time)