Resources
Material worth reading alongside the packages. None of this is built with EpiAware; it is the thinking and the tooling the work grows out of.
The case for composable models
Why modelling evidence has to be timely, rigorous, and collaborative, why chaining models together or building one monolithic model each give something up, and what a composable alternative would look like.
- A composable ecosystem for infectious disease modelling, the argument in short.
We have written this up at length, and will link it here once it is public.
Modelling workflow
- A workflow for infectious disease modelling, a workflow for building and criticising these models, adapted to the realities of the field. Modularisation is one of its stages, and it is the workflow the composable components are designed to support.
- Bayesian Workflow, the general modelling workflow these tools sit inside.
The R ecosystem
EpiAware brings this work to Julia. The R packages remain the more mature route for most real-time analysis today.
- EpiNow2, reproduction numbers, growth rates, and nowcasts from delayed, right-truncated data.
- epinowcast, nowcasting incident cases, and the partner community behind much of this work.
- primarycensored, delay distributions with primary event censoring.
- scoringutils, scoring and comparing probabilistic forecasts.
See epinowcast & the forum for more on the partner community.
Other Julia tools
Other efforts working towards composable infectious disease modelling, built outside EpiAware. They are listed with the packages, on the Packages page.
Learning Julia
New to the language? Start with the Using Julia guide.