Generate a line list from a branching process model parametrised by reproduction number

Usage

sim_branching_process(
  changes = dplyr::tibble(t = c(0, 40), rt = c(2.5, 0.8)),
  max_time = 80,
  seed = Sys.time(),
  fn_Rt = cfg_step_fn(changes),
  fn_ip = ~ggoutbreak::test_ip,
  fn_kappa = ~1,
  imports_df = NULL,
  fn_imports = ~ifelse(.x == 0, 30, 0),
  fn_list_next_gen = list(),
  ...
)

Arguments

changes: a dataframe containing a t time column and R reproduction number parameter. This parameter is optional if fn_Rt is specified
max_time: maximum duration of simulation
seed: random seed
fn_Rt: can be specified instead of changes df. This is a vectorised function that accepts a time parameter and returns a reproduction number. If both this and changes are specified this takes preference.
fn_ip: a function that takes input vector t (and/or class) and returns an infectivity profile at times t.
fn_kappa: a vectorised function taking t and other imported case metadata returning a dispersion parameter controlling the likelihood of individual super-spreading. This must be between 1 and Inf with 1 being standard poisson dispersion and larger values representing over dispersion.
imports_df: a data frame containing minimally time and count columns plus any metadata about the imports in additional columns. Metadata columns can inform the fn_Rt,fn_kappa and fn_ip functions as additional parameters.
fn_imports: a time varying function the defines the number of infected importations. If imports_df is defined then this is used instead
fn_list_next_gen: a named list of functions. The name corresponds to metadata columns in the simulation, the function is a purrr style mapping that will replace the old value in the named column with a new one. Such a function can be generated with cfg_transition_fn() when a transition probability matrix is involved, of it can be specified directly as a case_when style function. The function must be vectorised and assume no grouping structure. If the function has named parameters it can reference any of the metadata columns, or time (as t). The rcategorical() function may be useful in this scenario.
...: not used

Value

a line list of cases for a simulated outbreak

A dataframe containing the following columns:

id (unique_id) - Patient level unique id
time (ggoutbreak::time_period) - Time of infection. A `time_period`

Any grouping possible.

Examples

tmp = sim_branching_process(
  changes = dplyr::tibble(t = c(0,40), R = c(1.5,0.8)),
  max_time = 120,
  seed = 100,
  fn_imports = ~ ifelse(.x<10,1,0)
)
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> .
#> complete

if(interactive()) {
  plot_cases(tmp, mapping=ggplot2::aes(fill=as.factor(generation)),linewidth=0.1)
}

# imports can also be specified as a dataframe, which allows additional
# metadata in the line list. An example of which is as follows:
imports_df = dplyr::tribble(
  ~time, ~variant, ~count,
  0:4, "wild-type", 100,
  10:14, "alpha", 5,
)