Table 4 Prioritization scheme framework featuring the list of steps required to create final prioritization maps

1

Load relevant R packages and libraries

2

Obtain administrative boundaries of country/region of interest by either loading a pre-existing shapefile or directly through GADM [33]

3

Obtain water boundary shapefile layers from OCHA Humanitarian Data Exchange [34] (if relevant)

4

Obtain temperature suitability index raster layer from Malaria Atlas Project [5, 35]. If a temperature suitability index raster is not available or accessible, a combination of mean monthly rainfall, land surface temperature, and elevation can be used

5

Obtain monthly mean rainfall raster layers from CHIRPS [6]

6

Import spatially interpolated raster layer featuring proportion of households with at least 1 ITN per 2 people (if available). This is created using cluster-level DHS or MIS data [24] pertaining to ITN use and raster layers of covariates [5, 35] such as travel time to nearest city of 60,000 or more inhabitants, 2015 educational attainment for women of reproductive age, 2015 prevalence of improved housing, and the 2015 settlement model (a combination of built environment and population density), via the INLA-SPDE package

7

Aggregate relevant indicators (see Table 1 for examples) to desired administrative boundary level and add to a master dataset that is linked to the name and unique code for the desired administrative boundary. The unit which data is aggregated to is dependent upon the resolution of intervention data available (for example number of persons receiving SMC per local government area). Ensure that desired administrative boundary shapefile exists prior to aggregating values to desired administrative level

8

Convert categorical variable to most appropriate categorical variables from literature review and/or use natural breaks from ‘getJenksBreaks’ function in R with desired number of classes

9

Assign rank values to indicators, keeping in mind that the complement (or inverse) of some indicators may need to be calculated in order to maintain consistency with increased or decreased risk scores. For example, increased rainfall values (larger positive value) may imply increased malaria risk while, high population density, high urbanization (larger positive value) may imply decreased malaria risk especially if the main vector’s preferred larval habitats are in rural settings

10

Obtain factor-specific weights through application of the AHP or Delphi method. See GitHub repository [31] for syntax on calculating AHP weights using the ‘ahpsurvey’ package in R. Multiply weight value (as featured in Table 1) by rank values for each factor to create factor-specific score

11

Summate final factor-specific risk scores for each administrative unit so that each administrative boundary has cumulative prioritization score where a higher total score is indicative of higher prioritization for ITN targeting