DEN.spatial.RCT: Simulate a randomised control trial of dengue prophylactic...

In obrady/SpatialDengue: Fits and simulates a spatial model of the vector-borne disease dengue

Description

Provide basic information on the RCT and the number of stochastic runs over which model runs should be averaged

Usage

DEN.spatial.RCT(weekdates, fitdat, pastdat, unipix, pixdistmat, RCTinfo,
  nruns, paramsList = NULL)

Arguments

weekdates	Two element vector of the start and end weeks of the simulation over which the mdoel will be evaluated over
fitdat	Data frame of the locations, numbers and timings (in weeks) of cases to fit the model to, see ?sgdat
pastdat	Data frame of the locations, numbers and timings (in weeks) of all cases in the dataset (is used to generate the starting immunity profile), see ?sgdat
unipix	Universal pixel lookup table, see ?make.unipix
pixdistmat	A patch distance matrix, see example
RCTinfo	A data frame including details on RCT start and end time, control and treatment patches and effective coverage reached in treatment patch
nruns	integer, number of stochastic runs of the model over which results should be averaged
paramsList	Optional parameter list. If not supplied returns to defaults, see tutorial for full parameter list, see ?model.run for full list and explanation of parameters

Details

runs "nruns" number of stochastic simulations and treats people in the trial treatment clusters at the specified time. Returns a named list of six matrices detailing the number of seroconversions and symptomatic cases in treatment and control patches with each columns of each matrix representing one realisation from the stochastic model. The final two elements in the list detail the total person-days of observation and total number of people treated with drugs in the trial.

Examples

data(sgdat)
data(sgpop)
sgpop <- pop.process(sgpop, agg = 10)
unipix <- make.unipix(sgpop)
pixdistmat <- distm(cbind(unipix$x, unipix$y))
sgdat <- data.frame(sgdat, patchID = apply(cbind(sgdat[, 3:2]), 1, pix.id.find, unipix))
weekdates <- c(40, 92)
potential = (1:nrow(unipix))[unipix$pop > 1000]
potentialpops = unipix$pop[unipix$pop > 1000]
potential = cbind(potential, potentialpops)
randSam = sample(potential[, 1], 100)
TreatLocs = randSam[51:100]
TreatLocspop = potential[potential[, 1] %in% randSam[51:100], 2]
ContLocs = randSam[1:50]
ContLocspop = potential[potential[, 1] %in% randSam[1:50], 2]

RCTinfo <- list(Tstart = 50,
                Tend = 54,
                TreatLocs = TreatLocs,
                ContLocs = ContLocs,
                EffectiveCoverage = 0.9)

denmod_RCT = DEN.spatial.RCT(weekdates, sgdat, sgdat, unipix, pixdistmat, RCTinfo, nruns = 10)

# inspect results
nruns = 10
DEfigs <- rep(NA, nruns)
for(i in 1:nruns){
     # attack rate untreated
     ARU <- (sum(denmod_RCT$Control_infections[, i]) / sum(ContLocspop))
     # attack rate treated
     ART <- (sum(denmod_RCT$Treat_infections[, i]) / sum(TreatLocspop))
     # calculate drug efficacy
     DE = 100 * (ARU - ART) / ARU
     DEfigs[i] = DE
}
# ! uncertainty = model uncertainty
# other areas of uncertainty to consider = 
# site randomization, cluster size and ratio, drug effective coverage, timing of trial 
boxplot(DEfigs, ylim = c(0, 100), main = "Drug Efficacy")
summary(DEfigs)

obrady/SpatialDengue documentation built on Nov. 27, 2020, 12:13 p.m.