Skip to contents

EpiEstim reproduction number

Source: R/estimator-rt-epiestim.R
rt_epiestim.Rd

Calculate a reproduction number estimate from incidence data using the EpiEstim library and an empirical generation time distribution. This uses resampling to transmit uncertainty in generation time estimates. This is quite slow for each time series depending on the number of bootstraps and samples in the infectivity profile.

Usage

rt_epiestim(
  df = i_incidence_input,
  ip = i_infectivity_profile,
  bootstraps = 2000,
  window = 14,
  mean_prior = 1,
  std_prior = 2,
  ...
)

Arguments

df

Count data. Extra groups are allowed.

A dataframe containing the following columns:

  • count (positive_integer) - Positive case counts associated with the specified timeframe

  • time (as.time_period + group_unique) - A (usually complete) set of singular observations per unit time as a `time_period`

Ungrouped.

No default value.

ip

infectivity profile

A dataframe containing the following columns:

  • boot (anything) - a bootstrap identifier

  • time (positive_double) - the end of the time period (in days)

  • probability (proportion) - the probability of infection between previous time period until `time`

Must be grouped by: boot (exactly).

A default value is defined.

bootstraps

  • the number of bootstraps to take to calculate for each point.

window

  • the width of the epiestim window

mean_prior

the prior for the $R_t$ estimate. When sample size is low the $R_t$ estimate will revert to this prior. In EpiEstim the default is a high number to allow detection of insufficient data but this tends to create anomalies in the early part of infection timeseries. A possible value is $R_0$ but in fact this also will be a poor choice for the value of $R_t$ when case numbers drop to a low value.

std_prior

the prior for the $R_t$ SD.

...

not used

Value

A dataframe containing the following columns:

  • time (as.time_period + group_unique) - A (usually complete) set of singular observations per unit time as a time_period

  • rt.fit (double) - an estimate of the reproduction number

  • rt.se.fit (double) - the standard error of the reproduction number

  • rt.0.025 (double) - lower confidence limit of the reproduction number

  • rt.0.5 (double) - median estimate of the reproduction number

  • rt.0.975 (double) - upper confidence limit of the reproduction number

No mandatory groupings.

No default value.

Details

This will calculate a reproduction number for each group in the input dataframe.

Examples

tmp = growthrates::england_covid %>%
  time_aggregate(count=sum(count))

if (FALSE) {
  # not run due to long running
  tmp2 = tmp %>% rt_epiestim()
}