EpiNow2 (version 1.7.1)

forecast_infections: Forecast infections from a given fit and trajectory of the time-varying reproduction number

Description

This function simulates infections using an existing fit to observed cases but with a modified time-varying reproduction number. This can be used to explore forecast models or past counterfactuals. Simulations can be run in parallel using future::plan().

Usage

forecast_infections(
  estimates,
  R = NULL,
  model = NULL,
  samples = NULL,
  batch_size = 10,
  backend = "rstan",
  verbose = interactive()
)

Value

A list of output as returned by estimate_infections() but based on results from the specified scenario rather than fitting.

Arguments

estimates

The estimates element of an epinow() run that has been done with output = "fit", or the result of estimate_infections() with return_fit set to TRUE.

R

A numeric vector of reproduction numbers; these will overwrite the reproduction numbers contained in estimates, except elements set to NA. Alternatively accepts a <data.frame> containing at least date and value (integer) variables and optionally sample. More (or fewer) days than in the original fit can be simulated.

model

A compiled stan model as returned by rstan::stan_model().

samples

Numeric, number of posterior samples to simulate from. The default is to use all samples in the estimates input.

batch_size

Numeric, defaults to 10. Size of batches in which to simulate. May decrease run times due to reduced IO costs but this is still being evaluated. If set to NULL then all simulations are done at once.

backend

Character string indicating the backend to use for fitting stan models. Supported arguments are "rstan" (default) or "cmdstanr".

verbose

Logical defaults to interactive(). If the progressr package is available, a progress bar will be shown.

See Also

generation_time_opts() delay_opts() rt_opts() estimate_infections() trunc_opts() stan_opts() obs_opts() gp_opts()

Examples

# \donttest{
# set number of cores to use
old_opts <- options()
options(mc.cores = ifelse(interactive(), 4, 1))

# get example case counts
reported_cases <- example_confirmed[1:50]

# fit model to data to recover Rt estimates
est <- estimate_infections(reported_cases,
  generation_time = generation_time_opts(example_generation_time),
  delays = delay_opts(example_incubation_period + example_reporting_delay),
  rt = rt_opts(prior = LogNormal(mean = 2, sd = 0.1), rw = 7),
  obs = obs_opts(scale = Normal(mean = 0.1, sd = 0.01)),
  gp = NULL,
  forecast = forecast_opts(horizon = 0)
)

# update Rt trajectory and simulate new infections using it
R <- c(rep(NA_real_, 26), rep(0.5, 10), rep(0.8, 14))
sims <- forecast_infections(est, R)
plot(sims)

# with a data.frame input of samples
R_dt <- data.frame(
  date = seq(
    min(summary(est, type = "parameters", param = "R")$date),
    by = "day", length.out = length(R)
  ),
  value = R
)
sims <- forecast_infections(est, R_dt)
plot(sims)

#' # with a data.frame input of samples
R_samples <- summary(est, type = "samples", param = "R")
R_samples <- R_samples[
  ,
  .(date, sample, value)
][sample <= 1000][date <= "2020-04-10"]
R_samples <- R_samples[date >= "2020-04-01", value := 1.1]
sims <- forecast_infections(est, R_samples)
plot(sims)

options(old_opts)
# }