Apply a right censoring to count data.

Delayed observations means that if, for example a case is only attributed to a disease after a delay there is right censoring of the data. There can be very complex patterns of right censoring if for example observations are batched and published weekly. During COVID in the UK some death data was published frequently but some was retrospectively reported on monthly intervals, depending on where the patient died, which lead to complex time dependent biases in the death data. Given the description of the delay, this function will simulate this effect for count data. In another example there were delays reporting test results in the run up to Christmas which resulted in case rates apparently dropping as schools broke up. This could have affected timing of the 2021 Christmas lockdown.

Usage

sim_delayed_observation(
  df = i_sim_count_data,
  delay_fn,
  ...,
  input = "infections",
  output = input,
  max_time = max(df$time)
)

Arguments

df

a count dataframe from e.g. sim_poisson_model() or sim_summarise_linelist()

delay_fn

a function that takes time and returns the probability of observation (given it occurred) over time since infection (i.e. tau) as an ip delay distribution. This does not have to sum to 1 (e.g. mapping incidence to prevalence) but if not then it will behave as if some fraction or events are not observed (or observed multiple times). See cfg_weekly_ip_fn() and cfg_gamma_ip_fn() for helper functions to construct this parameter.

...

not used

input

the input statistic (defaults to count)

output

the output column name (defaults to same as input)

max_time

the date on which censoring is taking place.

Value

the result of applying this right censoring to the data.

Examples

weekday_delay = make_gamma_ip(median_of_mean = 5, median_of_sd = 2)
weekend_delay = make_gamma_ip(median_of_mean = 7, median_of_sd = 2)

delay_fn = ~ ifelse(.x %% 7 %in% c(6,7), list(weekend_delay), list(weekday_delay))

data = tibble::tibble(time=1:40, count = rep(100,40), statistic="infections") %>%
  dplyr::group_by(statistic) %>%
  sim_delayed_observation(delay_fn,output="delayed")

if (interactive()) ggplot2::ggplot(data,ggplot2::aes(x=time,colour=statistic))+
  ggplot2::geom_line(ggplot2::aes(y=count))