simulate_outbreak: Simulate outbreaks with transmission trees
In reconhub/simulacr: Outbreak simulation using branching processes

Description Usage Arguments Details Author(s) Examples

View source: R/simulate_outbreak.R

This function implements an individual-based outbreak simulator which generates transmission trees. A Poisson branching process is used to generate new cases in time, using the distribution of the reproduction number (R) and of the duration of the infectious period to determine rates of infection. A density-dependence term is used so that individual infectiousness decreases with the proportion of susceptible individuals in the population (see details). For each simulated case, the simulator generates an individual reproduction number, date of infection, symptom onset, and reporting using user-specified distributions. Distributions can be provided either as distcrete objects, as functions computing the probability mass functions (PMF), or as vectors of numbers taken as the PMF for values on 0, 1, ..., length(input) - 1

simulate_outbreak(
  duration = 100,
  population_size = 100,
  R_values,
  dist_incubation,
  dist_infectious_period,
  dist_reporting = NULL
)

`duration`	the number of days to run the simulation for
`population_size`	the number of susceptible hosts to use in the simulation; defaults to 100
`R_values`	a vector of values to be used as reproduction number; values will be drawn at random from this vector to determine the expected numbero of secondary cases for each new case
`dist_incubation`	the distribution of the incubation period, i.e. the time interval between infection and onset of symtpoms
`dist_infectious_period`	the distribution of the infectious period, i.e. the time interval between the moment a case starts showing symptoms (onset) and the moment they infect new secondary cases
`dist_reporting`	(optional) the distribution of the reporting delay, i.e. the time interval between symptom onset and the date at which the case is notified; if `NULL` (default) reporting dates will not be simulated

Individual infectiousness is determined as R x w(t - t_{onset}) where R is the individual reproduction number, w is the PMF of the duration of the infectious period, t is current time, and t_{onset} is the date of symptom onset for the considered individual. The number of new cases at time t is then taken from a Poisson distribution with a rate of infection λ_t n_s / n where lambda_t is the sum of all individual infectiousness at time t, n_s is the number of susceptible individuals in the population, and n is the total population size.

Thibaut Jombart thibautjombart@gmail.com

## make toy distributions (showing the 3 possible input types)
incubation <- c(0, 1, 1, 1, 1) # numbers = unscaled PMF
infectious_period <- make_disc_gamma(10, 7) # distcrete object
reporting <- function(x) dpois(x, 5) # PMF function
set.seed(1)
x <- simulate_outbreak(R = runif(100, 1, 3), # random values on [1;3]
                       dist_incubation= incubation,
                       dist_infectious_period = infectious_period,
                       dist_reporting = reporting)
dim(x)
head(x)
tail(x)
if (require(epicontacts)) {
  plot(x)
}