adjust_infection_to_report: Adjust from Case Counts by Infection Date to Date of Report

In epiforecasts/EpiNow: Estimate Realtime Case Counts and Time-varying Epidemiological Parameters

Description

Stochastic mapping from cases by date of infection to date of report via date of onset. Essentially reversal of nowcast_pipeline.

Usage

adjust_infection_to_report(
  infections,
  delay_def,
  incubation_def,
  reporting_effect,
  reporting_model,
  type = "sample",
  return_onset = FALSE,
  truncate_future = TRUE
)

Arguments

infections	data.table containing a date variable and a numeric cases variable.
delay_def	A single row data.table that defines the delay distribution (model, parameters and maximum delay for each model). See lognorm_dist_def for an example of the structure.
incubation_def	A single row data.table that defines the incubation distribution (model, parameters and maximum delay for each model). See lognorm_dist_def for an example of the structure.
reporting_effect	A numeric vector of length 7 that allows the scaling of reported cases by the day on which they report (1 = Monday, 7 = Sunday). By default no scaling occurs.
reporting_model	A function that takes a single numeric vector as an argument and returns a single numeric vector. Can be used to apply stochastic reporting effects. See the examples for details.
type	Character string indicating the method to use to transfrom counts. Supports either "sample" which approximates sampling or "median" would shift by the median of the distribution.
return_onset	Logical, defaults to FALSE. Should cases by date of onset also be returned?
truncate_future	Logical, should cases be truncted if they occur after the first date reported in the data. Defaults to TRUE.

Value

A data.table containing a date variable (date of report) and a cases variable. If return_onset = TRUE there will be a third variable reference which indicates what the date variable refers to.

Examples

## Define example cases
cases <- data.table::as.data.table(EpiSoon::example_obs_cases) 

cases <- cases[, `:=`(confirm = as.integer(cases), import_status = "local")]

## Define a single report delay distribution
delay_def <- EpiNow::lognorm_dist_def(mean = 5, 
                                      mean_sd = 1,
                                      sd = 3,
                                      sd_sd = 1,
                                      max_value = 30,
                                      samples = 1,
                                      to_log = TRUE)
                                       
## Define a single incubation period
incubation_def <- EpiNow::lognorm_dist_def(mean = EpiNow::covid_incubation_period[1, ]$mean,
                                           mean_sd = EpiNow::covid_incubation_period[1, ]$mean_sd,
                                           sd = EpiNow::covid_incubation_period[1, ]$sd,
                                           sd_sd = EpiNow::covid_incubation_period[1, ]$sd_sd,
                                           max_value = 30, samples = 1)
                                           

## Perform a nowcast
nowcast <- nowcast_pipeline(reported_cases = cases, 
                            target_date = max(cases$date),
                            delay_defs = delay_def,
                            incubation_defs = incubation_def)
                            

infections <- nowcast[type %in% "infection_upscaled" & import_status %in% "local"]
infections <- infections[, `:=`(type = NULL, import_status = NULL)]


## Simple mapping
report <- adjust_infection_to_report(infections, delay_def, incubation_def)   

print(report)   

## Mapping with a weekly reporting effect
report_weekly <- adjust_infection_to_report(
                      infections, delay_def, incubation_def,
                      reporting_effect = c(1.1, rep(1, 4), 0.95, 0.95))          
                             
print(report_weekly) 

## Map using a deterministic median shift for both delays
report_median <- adjust_infection_to_report(infections, delay_def, 
                                           incubation_def, type = "median")      
                                           
                                                   
                                                           
## Map with a weekly reporting effect and stochastic reporting model
report_stochastic <- adjust_infection_to_report(
                      infections, delay_def, incubation_def,
                      reporting_effect = c(1.1, rep(1, 4), 0.95, 0.95),
                      reporting_model = function(n) {
                      out <- suppressWarnings(rnbinom(length(n), as.integer(n), 0.5))
                      out <- ifelse(is.na(out), 0, out)
                      })          
                             
print(report_stochastic)         

epiforecasts/EpiNow documentation built on Oct. 26, 2020, 2:38 p.m.