Disagapp: A Shiny app to facilitate reproducible disaggregation regression analyses

Creating high-resolution maps of disease risk is important for many climatically-driven diseases including vector-borne, zoonotic and non-communicable diseases. However, disease counts are commonly only available aggregated to an administrative level such as the county, department or province. Using disease mapping to make high-resolution risk maps from these data can be challenging, even though high-resolution data on temperature and other environmental variables are available. Disaggregation regression is a new multiscale modelling framework that has been used in mapping global malaria incidence and dengue incidence, amongst others. We present Disagapp (https://github.com/simon-smart88/disagapp; https://disagapp.le.ac.uk/) a web app for disease mapping with administrative-level data. The app was designed to be user-friendly, with a point and click interface and with modelling guidance integrated at each step of the analysis. It can be accessed online or run locally by installing the R package and calling one function. Environmental and economic covariates (temperature, precipitation, distance to water, land use, population density, accessibility and night lights) are seamlessly retrieved from various sources, collated and harmonised and used to fit disaggregation regression models using the disaggregation R package. This modelling framework is a principled way to create high-resolution predictions of disease risk. Analyses conducted in the app can be reproduced outside of the app by generating an R markdown document, allowing the user to share, preserve, extend or learn from their analysis. By making the techniques available in the disaggregation package available online and making it simple and easy to access covariates, we remove barriers for non- or novice R users or analysts who work in institutions with restrictive installation permissions to access these innovations. Therefore, this new app makes this important modelling framework much more available to the global modelling community.