diseasy: quick start guide

library(diseasy)
#> Loading required package: diseasystore
#> 
#> Attaching package: 'diseasy'
#> The following object is masked from 'package:diseasystore':
#> 
#>     diseasyoption

Overview

The diseasy package is large in scope as it aims to provide the infrastructure to easily create a large ensemble of disease models.

This building of disease models is achieved via a modular design philosophy. To that end, we have a number of “functional” modules that feed into “model” modules.

Functional modules

The functional modules are smaller modules meant for defined purposes. These modules are “atomic” and form the building blocks of the ensembles.

The currently available functional modules are:

ls("package:diseasy") |>
  purrr::keep(~ stringr::str_detect(., pattern = "Diseasy(?!(BaseModule|Model.*))"))
#> [1] "DiseasyActivity"    "DiseasyObservables" "DiseasySeason"     
#> [4] "DiseasyVariant"

Each functional module is restricted to covering a single functionality for use in the models (and sometimes for other functional modules).

For example, DiseasySeason defines a number of models for the season dependency of the disease. This module has a single defined purpose (modelling the effect of season) but it provides a number of different models of describing possible season dependencies:

s <- DiseasySeason$new()
s$available_season_models
#> [1] "constant_season" "cosine_season"   "covid_season_v1" "covid_season_v2"

As we will see later, these different models for season dependency will be used to construct the ensembles.

Each of these season models have similar functional signatures (i.e. the function calls to the models are similar). This way, when we create a disease model, we can use the module to implement different models for season dependency in the disease models easily.

The other modules have other functionality they implement. For example, the module DiseasyObservables is responsible for providing data to the models and modules. For DiseasySeason, this means that some models of season dependency can use real-world data, such as temperature, in the season models.

The use of these modules is varied and each module comes with their own vignette describing their usage: vignette("DiseasyActivity"), vignette("DiseasyObservables"), vignette("DiseasySeason"), vignette("DiseasyVariant").

Model modules

As stated above, the functional modules provide functionality for the disease models.

The currently available model modules are:

ls("package:diseasy", pattern = "DiseasyModel.+")
#> character(0)

Presently, no models are included in the package. These will come in the following releases.

Configuring

All diseasy modules are build on top of the DiseasyBaseModule class. This class provides a number of features that are common to all modules, such as logging and caching.

Caching

Internally, the modules may use cachem to cache computations either in memory or on disk. To control how this caching is done, you can provide the cache argument to the module constructor containing a cachem cache object, or you can set the option diseasy.cache to a cachem cache object. See DiseasyBaseModule for more information.