whoforecast

Authors: Martina McMenamin

Preamble

This is an experimental package to run nowcasting and forecasting analysis and create a report output. Please note that the user of this package is responsible for the outputs and these are not WHO endorsed.

THIS APP IS CURRENTLY IN EARLY DEVELOPMENT - PLEASE LOG A GITHUB ISSUE OR CONTACT THE AUTHORS FOR ANY ISSUES. ADDITIONAL FUNCTIONALITY FOR SUBNATIONAL ESTIMATES AND MORE DETAILS IN OUTPUT REPORT ARE IN THE PROCESS OF BEING ADDED.

Installation

To install this package for the first time run the code below. The package is at the early stages of development so will need to be reinstalled at regular intervals to get the most up to date features.

devtools::install_github("whocov/whoforecast", 
                          force = TRUE, 
                          dependencies = TRUE)

Prepare data

To use this package, you will need to have a dataset with the following columns:

• date: the date of the data point
• cases: the number of cases on that date
• iso_3_code: the ISO 3 code of the country
• adm0_name: the name of the country or adm1_name if regional level analysis

Run analysis

To run the analysis, you will need to follow the steps below:

library(whoforecast)

# Load the data 
test_data <- readxl::read_csv("/your_path/example_data.csv") 

# Run the analysis - edit the function arguments as needed
run_forecast(
    data = test_data, # read in your data
    adm_level = "adm0", # set to "adm0" if national level and "adm1" if regional level analysis
    reporting_freq = "weekly", # set to "weekly" or "daily" based on the reporting frequency in the data
    generation_time = get_gen_time(), # set to the generation time for the disease - uses default but can be specified
    incubation_period = get_inc_period(), # set to the incubation period for the disease - uses default but can be specified
    reporting_delay = get_rep_delay(), # set to the reporting delay for the disease - uses default but can be specified
    forecast = TRUE, # set to TRUE if you want to include forecasting analysis
    horizon = 7, # set to the number of days you want to forecast
    create_report = TRUE, # set to TRUE if you want to create a report in Word
    date_from = min(test_data$date), # set to the start date for the estimation
    date_var = "date", # set to the name of the date column in the data
    case_var = "cases" # set to the name of the cases column in the data
    )

This will run the national level analysis and output both nowcasted and forecasted estimates, with forecast period according to horizon. If a subnational level analysis is run, change adm_level = "adm1".

Analysis output

The analysis will output a list with the following elements of interest:

• Model estimates
• Figure of estimates for effective reproduction number (Rt) over time - this is a modified version of the figure output by EpiNow2 package to allow for aggregate data
• Figure of estimates for the number of reported cases over time - this is a modified version of the figure output by EpiNow2 package to allow for aggregate data

If create_report = TRUE a Word document with the analysis output and interpretation will be created and will open automatically when the analysis has finished running. Note if you wish to run the analysis again you must close the Word document - otherwise an error will be shown.

Disclaimer

• If nowcasting is included, the accuracy of this will depend on modifying the input reporting delay distribution estimated from timestamped datasets
• The forecasts do not account for interventions or changes in other factors during this period
• Development of report is ongoing to include presentation of lower admin levels