This repository contains scripts used in the analysis presented in the manuscript titled Enhancing Mass Vaccination Programs with Queueing Theory and Spatial Optimization. A description of the repository's contents are provided below.
Scripts used to optimize the locations of vaccination sites via the hybrid recursive interchange-genetic algorithm are located in the scripts_optimization folder. A description of these scripts are provided below:
- functions_recursiveinterchange.R - contains all the functions needed to run the recursive interchange component of the hybrid algorithm
- functions_geneticalgorithm.R - contains all the functions needed to run the genetic algorithm component of the hybrid algorithm
- optimize...R - each script file runs a single round of the hybrid algorithm
- The values of alpha and beta are specified in the portion of the script name that comes after the underscore "_" (alpha = beta = 0 are specified as "queuenaive")
- The round of the optimization is given by the number following optimize
- So as an example, optimize2_alpha0.1beta0.1.R was used to run the second round of optimization for the high-attrition scenario (alpha = beta = 0.1)
- check_stoppingcondition.R - determines at the conclusion of each round of optimization whether the stopping condition had been met
Scripts used to run the queueing simulations, generate figures, and conduct the sensitivity analyses are located in the scripts_other folder. A description of these scripts are provided below:
- simulate_vsequilibriumsolutions.R - compares results of the stochastic queueing model to the closed-form equations
- figure_simulations.R - runs single trials of the queueing model and plots the results
- figure_potentialvaxsites.R - maps the locations of all potential vaccination sites
- figure_arrivalhistograms.R - generates the arrivals histograms that were presented in the manuscript
- figure_catchmentmaps.R - maps the locations of optimized vaccination sites and numbers of expected arrivals that were presented in the manuscript
- sfig_poissonregression.R - fits and plots the MDVC participation probability function and the post-MDVC survey data that was used in the regression
- sa_balkingreneging.R - evaluates sensitivity of results to misspecification of alpha and beta
- sa_inconstantarrivals....R - evaluates sensitivity of results to the assumption of a constant arrival rate
Non-sensitive data used in the analysis are provided as a single RDA file, which can be found in the data folder. Note that the locations of houses in the study area and the probability-distance matrix (containing the probability of each household participating in the campaign for all possible vaccination sites) have been removed to protect privacy.
- optimization_data.rda
- asa: shapefile of Alto Selva Algre (study area)
- queuedata: times of arrivals collected during an MDVC
- supplycoord: coordinates of the potential vaccination sites in UTM projection
- survey: survey data used to fit the MDVC participation probability function