Maplaria: a user friendly web-application for spatio-temporal malaria prevalence mapping

Background

Model-basedgeostatistical (MBG) methods have been extensively used to map malaria riskusing community survey data in low-resource settings where disease registriesare incomplete or non-existent. However, the wider adoption of MBG methods bynational control programmes to inform health policy decisions is hindered bythe lack of advanced statistical expertise and suitable computationalequipment. Here, Maplaria, an interactive, user-friendlyweb-application that allows users to upload their own malaria prevalence dataand carry out geostatistical prediction of annual malaria prevalence at anydesired spatial scale, is introduced.

Methods

In thedesign of the Maplaria web application, two main criteria were considered: theapplication should be able to classify subnational divisions into the mostlikely endemicity levels; the web application should allow only minimal inputfrom the user in the set-up of the geostatistical inference process. To achievethis, the process of fitting and validating the geostatistical models iscarried out by statistical experts using publicly available malaria survey datafrom the Harvard database. The stage of geostatistical prediction is entirelyuser-driven and allows the user to upload malaria data, as well as vector datathat define the administrative boundaries for the generation of spatiallyaggregated inferences.

Results

The processof data uploading and processing is split into a series of steps spread acrossscreens through the progressive disclosure technique that prevents the userbeing immediately overwhelmed by the length of the form. Each of these isillustrated using a data set from the Malaria Indicator carried out in Tanzaniain 2017 as an example.

Conclusions

Maplariaapplication provides a user-friendly solution to the problem makinggeostatistical methods more accessible to users that have not undertaken formaltraining in statistics. The application is a useful tool that can be used tofoster ownership, among policy makers, of disease risk maps and promote betteruse of data for decision-making in low resource settings.