WalkingDead2

The goal of WalkingDead2 is to simulate the events of infection (of susceptible subjects) and immunization (of infected subjects) in a population of susceptible, infected, and immunized individuals. It is an extension of the WalkingDead project by George Vegayon.

Installation

WalkingDead2 is available through the author’s Github (click here). In R, you can install and load the package using the following code:

# Installing WalkingDead2 from github 
devtools::install_github("Daniel-K-Addo/WalkingDead2")

Brief Details

Currently, WalkingDead2 accommodates the transition of subjects from the susceptible to infected state upon contact with an infected subject, as well as the transition of subjects from the infected to immunized state upon contact with an immunized subject. Immunized subjects do not undergo any transitions. This is practical for a communicable disease for which there is a cure that immunizes the infected but no vaccine for the susceptible. This scenario can very well be extended to different fields.

Functions and Arguments

There are eight arguments in, wd2, the primary function of WalkingDead2. Five of these have default values. Briefly, these arguments perform the following roles:

• n, m, k n represents the total population size; m represents the infected population size; k represents the immunized population size. Note: $n \geq 3$ and ${m,k} > 0$.
• fixedAllocation a logical which determines whether the initial sub population sizes are randomly determined using multinomial distribution OR fixed given {n,m,k}.
• infect numeric value between $0$ and $1$ representing the infection rate.
• infect_radius numeric value between $0$ and $1$ representing the radius of a transition-causing subject within which an eligible subject will undergo a status change when exposed.
• nIterations represents a number of runs during which each subject’s sub population membership and location will potentially be updated. This may be viewed as some unit of time.
• flight a logical which determines if the infected population are programmed to move away from the immunized population (flight) OR move towards the susceptible (fighting to keep population numbers up). By default, the susceptible population will always move away from the infected population whereas the immunized population will always move towards the infected population.

The output from wd2 has class $wd2$ and may be passed through print, summary, plot, and snapshot functions for varied results as demonstrated in the example below.

Example

This is a basic example which shows you how to solve a common problem:

# Load the package
library(WalkingDead2)
set.seed(2608)
# Primary function
temp <- wd2(n = 100, # Total Population size 
            m = 16, # Initial Infected Population size
            k = 4, # Initial Immunized Population size
            fixedAllocation = FALSE, # Allocation type: Randomize if FALSE
            infect = 0.75, 
            infect_radius = 0.075, # Transition-causing radius
            nIterations = 100, # Total number of runs
            flight = FALSE # Infected will flee from immunized if TRUE
            )

# Demonstrate print.wd2
print(temp) # or print.wd2(temp)
#> [1] "The pathogen was eliminated successfully after run 76"
#> $snapshot
#>        [,1] [,2] [,3]
#>  [96,]   33    0   67
#>  [97,]   33    0   67
#>  [98,]   33    0   67
#>  [99,]   33    0   67
#> [100,]   33    0   67
#> [101,]   33    0   67

# Demonstrate summary.wd2 and view first and last few rows
summary(temp) |> head(10) # or summary.wd2(temp)
#>     Run Susceptible Infected Immune ChangeSI ChangeII Transitions
#>  1:   0        0.83     0.12   0.05       NA       NA          NA
#>  2:   1        0.73     0.20   0.07       10        2          12
#>  3:   2        0.68     0.25   0.07        5        0           5
#>  4:   3        0.65     0.27   0.08        3        1           4
#>  5:   4        0.62     0.29   0.09        3        1           4
#>  6:   5        0.60     0.31   0.09        2        0           2
#>  7:   6        0.60     0.31   0.09        0        0           0
#>  8:   7        0.60     0.31   0.09        0        0           0
#>  9:   8        0.60     0.31   0.09        0        0           0
#> 10:   9        0.59     0.32   0.09        1        0           1
summary(temp) |> tail(5)
#>    Run Susceptible Infected Immune ChangeSI ChangeII Transitions
#> 1:  96        0.33        0   0.67        0        0           0
#> 2:  97        0.33        0   0.67        0        0           0
#> 3:  98        0.33        0   0.67        0        0           0
#> 4:  99        0.33        0   0.67        0        0           0
#> 5: 100        0.33        0   0.67        0        0           0

# Demonstrate plot.wd2
plot(temp) # or plot.wd2(temp)

# Demonstrate snapshot
snapshot(temp)
#>   SampleSize InfectionRate Flight Run PropSusceptible PropInfected
#> 1        100          0.75  FALSE  77            0.33            0
#>   PropRecovered Equilibrium
#> 1          0.67        TRUE

#> [1] "The pathogen was eliminated successfully after run 76"

This second plot shows the case when flight is set to TRUE.

#> [1] "The pathogen was NOT eliminated. Infections are still possible."

Simulations

It is possible to run multiple simulations with varying parameters of interests for different purposes. An example is demonstrated below where the interest could be in the number of times EQUILIBRIUM was acheived within a 100 trials. The example uses the advantage of parallel computing and the snapshot function the WalkingDead2 package. This example is not ran.

require(dplyr)
require(WalkingDead2)

NN <- seq(50, 200, length=3) # Sample size options
RR <- seq(0.25, 0.75, length=3) # Infection Rate options
FF <- c(TRUE, FALSE) # Flight or fight
NNRRFF <- expand.grid(NN,RR,FF) # Combinations of NN and RR
NNRRFFrep <- NNRRFF %>% slice(rep(1:n(), each=500))

simwd2 <- function(j){
  NN <- NNRRFFrep[j,1]
  RR <- NNRRFFrep[j,2]
  FF <- NNRRFFrep[j,3]
  testRun <- wd2(n = NN, # Population size
                 m = round(0.15*NN), # Initial Infected Population size
                 k = round(0.10*NN), # # Initial Immunized Population size
                 fixedAllocation = FALSE, # Allocation type: randomized, if FALSE
                 infect = RR, # infection rate
                 infect_radius = .075, # radius of infection
                 nIterations = 50, # Number of runs
                 flight = FF # Infected will flee from immunized if TRUE
  )
  keepThis <- snapshot(testRun)
  return(keepThis)
}

# using parallel package with at least 16 cores available
require(parallel)
cl <- parallel::makeCluster(16)
clusterExport(cl,c("NNRRFFrep", "simwd2"))#,"wd2","simwd2","snapshot"))
clusterEvalQ(cl, library(WalkingDead2))
temp <- do.call(rbind.data.frame,parLapply(cl,1:dim(NNRRFFrep)[1],simwd2))
stopCluster(cl)
saveRDS(temp, "SimWD2Out.rds")

# OR using slurmR on a high performance computer
require(slurmR)
cl <- makeSlurmCluster(16, partition = "notchpeak-shared-short",
                       account = "notchpeak-shared-short")
clusterExport(cl,c("NNRRFFrep", "simwd2"))#,"wd2","simwd2","snapshot"))
clusterEvalQ(cl, library(WalkingDead2))
temp <- do.call(rbind.data.frame,parLapply(cl,1:16,simwd2))
stopCluster(cl)
saveRDS(temp, "SimWD2Out.rds")