Worked Example 1 - Running Time Delay Model

This worked example demonstrates how to set up and run the time delay version of the standard yellow fever dynamic model presented in the YEP package. The time delay version differs from the standard version (which is described in this worked example) in using a fixed time delay to govern the movement of infected people from group E (exposed) to group I (infectious) and from group I to group R (recovered). In the standard model, this movement is governed by a rate given by the inverse of the time delay values (the sum of [TBA] in the case of E-I and the infectious period in the case of I-R). The [TBA] are all set to 5 days within the package code. [TBA - how to alter them]

Using a fixed delay period increases the complexity of the model (due to the properties of the architecture created using the odin.dust package) but is potentially more accurate when modelling yellow fever transmission over short timescales.

The inputs for the Model_Run_Delay() function, which runs this version of the model, are the same as for the Model_Run function in the YEP package, except that there is no variable output type - the function always outputs detailed SEIRV data for every time point and age group (equivalent to setting output_type=“full” in Model_Run()).

library(YellowFeverDynamics)
input_data <- readRDS(file = paste(path.package("YellowFeverDynamics"), 
                                   "/exdata/input_data_example.Rds", sep = ""))
vacc_data <- input_data$vacc_data[1, , ]
pop_data <- input_data$pop_data[1, , ]  
# Force of infection due to spillover from sylvatic reservoir
FOI_spillover <- 1.0e-8             
# Basic reproduction number for urban spread of infection
R0 <- 1.5             
# Year(s) for which data is to be output
years_data <- c(2000)
# SEIRV data from end of a previous run to use as input (if mode_start = 2; not 
# used here and hence set to NULL)
start_SEIRV <- NULL           
# First year in population/vaccination data (here taken from the input data set; 
# if the population and vaccination data is set manually, this similarly needs 
# to be set manually)
year0 <- input_data$years_labels[1]
# Flag indicating how to set initial population immunity level in addition to
# vaccination
# If mode_start = 0, only vaccinated individuals are immune (R = 0)
# If mode_start = 1, shift some non-vaccinated individuals into recovered to 
# give herd immunity (R = 1 - 1/R0)
# If mode_start = 2, use SEIRV input in list from previous run(s), set using
# start_SEIRV variable
mode_start <- 1  
# Proportional vaccine efficacy (from 0 to 1); probability of vaccination 
# successfully providing immunity (i.e. moving vaccinated individual to group V)
vaccine_efficacy <- 1.0       
# Time increment in days to use in model (should be 1.0, 2.5 or 5.0 days)
dt <- 1.0            
# Number of particles to run
n_particles <- 10     
# Number of threads to run
n_threads <- 1   
# TBA
deterministic <- FALSE

We run Model_Run_Delay() as shown:

test1 <- Model_Run_Delay(FOI_spillover, R0, vacc_data, pop_data, years_data,
                   start_SEIRV, year0, mode_start, vaccine_efficacy, 
                   dt, n_particles, n_threads, deterministic)

As with YEP::Model_Run(), we can plot the output:

library(YFDaux)
plot <- plot_model_output(test1)
print(plot)