Modeling the number of hospital admissions for malaria in South Africa by using climate variables as disease drivers

Recently we proposed a regression model for the number of hospital admissions for malaria in the Limpopo province of South Africa. We developed our model using the available weekly epidemiological reports from five districts in this province, in the period 2000-2020. We analyzed number of hospitalizations for malaria time series in relation to time series of temperature, rainfall and evaporation from bare soil ground or satellite data from the same geographical area and developed an algorithm that links combined changes in these three variables with the changes in number of malaria hospitalizations. We used wavelet spectral analysis to determine time lags in their cross-correlations.   We used this model to provide projections for the Limpopo malaria cases for the next five years (2025-2029). Since there are no future projections available for evapotranspiration, we used three different methods to estimate future values of this variable in our model: 1) a combination of temperature and rainfall data, 2) use of total soil moisture content records and their projections, and 3) use of Hargreaves empirical formula. We will present and compare our results for all three cases. Our calculations can be used for public health preparedness.