CamilleMorlighem/REACT2cities

Malaria risk mapping in sub-Saharan African cities # Malaria risk mapping in sub-Saharan African cities using environmental and socio-economic predictors This repository contains data and material to model and predict malaria risk (measured as *Pf*PR 2-10 ) in four sub-Saharan African cities, Dakar (Senegal), Ouagadougou (Burkina Faso), Kampala (Uganda) and Dar es Salaam (Tanzania), using a set of environmental and socio-economic predictors derived from remote sensing imagery. These predictors are multi-resolution variables depicting the urban climate, the land use and the land cover. *Pf*PR 2-10 modelling and prediction are achieved using a popular machine learning algorithm, namely random forest (RF). Malaria prevalence data come from various types of surveys, many of which are Demographic and Health Surveys (DHS) in which the urban cluster coordinates are randomly displaced within 2 km buffers to protect the privacy of the participants. Although it has little impact at the national scale, at the intra-urban scale it decreases the spatial accuracy of the DHS indicators and in turn affects the predictive performance of malaria models. The modelling workflow proposed here allows to use and test spatial optimisation methods to reduce the effect of DHS displacement on model predictive performance, as used 2-10 or --> in Georganos *et al.*, 2019 to model the DHS wealth index. These methods are based on the duplication of the released DHS coordinates in the four cardinal directions (N, S, E, W) in order to enrich or refine the contextual spatial feature extraction. Fore more information on the data and material presented here see Morlighem *et al.*, 2022. ## Data ### Malaria prevalence data Malaria prevalence data (see folder `Data\Malaria_data\`) come from an open online malaria database recording survey data from several sources from 1900 to 2015 [4], such as scientific papers or DHS. The prevalence is measured as the *Plasmodium falciparum* Parasite Rate (i.e. the proportion of people infected by *Pf*) sta …