SpatialSEIRModel: Fit a spatial/non-spatial SEIR/SEIRS model based on the...

In grantbrown/ABSEIR: Posterior Approximation to Fully Bayesian Spatial SEIR Models Using Multithreaded C++

Description

Fit a spatial/non-spatial SEIR/SEIRS model based on the provided model components.

Usage

SpatialSEIRModel(
  data_model,
  exposure_model,
  reinfection_model,
  distance_model,
  transition_priors,
  initial_value_container,
  sampling_control,
  samples = 100,
  verbose = FALSE,
  ...
)

Arguments

data_model	A data model object, describing the link between the observed data and the unobserved epidemic counts. Valid data models are created using the DataModel function.
exposure_model	An exposure model object, which describes the spatial and temporal variability of the exposure/infection process. Valid exposure models are created using the ExposureModel function.
reinfection_model	A reinfection model object, which describes whether or not individuals are able to return from the Removed category to the Susceptible population. Valid reinfection models are created using the ReinfectionModel function.
distance_model	A distance model object which describes the underlying contact network, in addition to prior parameters which constrain the contact process. Valid distance models are created using DistanceModel
transition_priors	An object containing information about the E to I and I to R transition prior parameters. These are created using the TransitionPriors function, and involve either transition probabilities and corresponding effective sample sizes (amount of prior information), or manually specified probability distributions. .
initial_value_container	An object specifying the initial state of the epidemic for each spatial location, created by the InitialValueContainer function.
sampling_control	An object specifying information about the sampling algorithm. In particular, the sampling_control argument should specify the number of CPU cores to employ, and the random seed to use. Sampling control objects are created by the SamplingControl function.
samples	the number of samples to approximate from the posterior distribution, i.e. the number of particles to simulate. The number of particles should be considerably smaller than the batch size specified by the sampling_control argument. Ignored for the debug-oriented 'simulate' algorithm.
verbose	print diagnostic information on the progress of the fitting algorithm. Available output levels are 0, 1, 2, and 3, in ascending order of detail. Level 0 output will print almost no progress/diagnostic information to the log. Level 1 Will provide iteration updates only. Leve 2 provides additional chain setup and diagnostic information. Level 3 prints calculation diagnostic information.
...	Additional arguments, used internally.

Details

Use the supplied model components to build and fit a corresponding model. This function is used to fit all of the models in the spatial SEIRS model class. Numerous ABC algorithms have been developed, but as of now ABSEIR provides just two. The first algorithm is the basic rejection algorithm of Rubin 1980. While this approach performs well when good prior information is available, it can be extremely inefficient when prior distributions are diffuse with respect to the posterior. To address this shortcoming, we have implemented the Sequential Monte-Carlo approach proposed by Beaumont 2009, 2010. We may provide additional algorithms in the future, in particular that of Del Moral et al. 2012.

Value

an object of type SpatialSEIRModel

See Also

DataModel, ExposureModel, ReinfectionModel, DistanceModel, TransitionPriors, InitialValueContainer, SamplingControl, summary.SpatialSEIRModel, plot.SpatialSEIRModel, compareModels, epidemic.simulations,

Examples

## Not run: results = SpatialSEIRModel(data_model, exposure_model,
                                                reinfection_model, distance_model,
                                                transition_priors, initial_value_container,
                                                sampling_control, 50, TRUE)
## End(Not run)

grantbrown/ABSEIR documentation built on Oct. 14, 2021, 2:32 p.m.