Generating spatial modeled estimates

⚠️ Section Under Development

Example use case of producing prevalence estimates

Draft outline (from Bea’s notes):

• Step 1: obtain prevalence estimates from the sources of data available (survey reports, literature, etc) through time - highlight the need to consider the sampling weights if this is going to be done manually (and offer some code examples)

  Ensure that prevalence data are available to define the “baseline” transmission (before community-based prevention interventions were scaled up).

• Step 2: Use spatio-temporal models to produce prevalence estimates for all units of analysis and periods of time possible.

  For this, the page should recommend working with an analysis partner (local or global) with these capacities and providing a list of considerations for the SNT team to focus on throughout this process (for example, review of covariates, review of results through interpretable outputs, etc)

  Consider providing some examples of packages (INLA) or documentation that independent analysts would like to use to ensure they are taking into consideration the main aspects for malaria transmission modeling (for ex: rainfall with a lag, elevation for mosquito presence, etc).

  Here MAP can be mentioned as a source of info, and ideally with contacts to the East Africa Node, explanations of their modeling approach, etc.

  We could also recommend and link to the mbg package if we like it.

• Step 3: Production of results

  If the geospatial modeled results are presented in rasters, then, how to use rasters to obtain results at the appropriate level (this is on the Working with Geospatial Modeled Estimates page).