Predicting malaria hyper endemic zones in West Africa using a regional scale dynamical malaria model

• Published in: Frontiers in Tropical Diseases
• Main Authors: Eniola A. Olaniyan, Adrian M. Tompkins, Cyril Caminade
• Format: Article in Journal/Newspaper
• Language: English
• Published: Frontiers Media S.A. 2024
• Subjects: ECMWF-seasonal-forecasts; malaria; early warning system; VECTRI; West Africa; Arctic medicine. Tropical medicine; RC955-962; Arctic
• Online Access: https://doi.org/10.3389/fitd.2024.1322502; https://doaj.org/article/a070fb5680964892b4d0160e33fec1e3

Due to the continuing large number of malaria-related deaths in tropical Africa, the need to develop a robust Malaria Early Warning System (MEWS) for effective action is growing to guide cost-effective implementation of interventions. This study employs a two-stage hierarchical evaluation technique to evaluate the ability of the VECTRI malaria model to simulate malaria dynamics at seasonal time scale (1 - 7 months) over Nigeria and West Africa. Two sets of malaria simulations are considered. The first set is based on VECTRI simulations driven by observed rainfall and temperature datasets (hereafter referred to as control run). The second is based on malaria simulations driven by the European Centre for Medium-Range Weather Forecasting (ECMWF) System5 ensemble seasonal forecasting system (hereafter referred to as Forecast run). Different metrics are employed to assess the skill of the VECTRI malaria model. Results based on the control run indicate that the model can reproduce hyper-endemic zones and the evolution of malaria cases, particularly the observed increase in cases with decreasing population density. Despite having significant biases and low correlation, the model successfully predicts annual anomalies in malaria cases across Nigeria, particularly in the savannah region that experience large malaria burden. Annual correlations between the VECTRI Forecast run and the VECTRI Control run are relatively low at all lead times (LT) and for each start date (SD) across West Africa, although correlation generally increases from the Gulf of Guinea to the Sahel. Despite low correlations, the Rank Probability Skill Score (RPSS) reveals that the model has a statistically significant skill in predicting malaria occurrences across all categories of malaria cases, regardless of start date or lead time. While the Guinea Forest has the strongest RPSS, the increase or decrease in skill from the first to seventh lead time varies significantly across the region. In addition, the VECTRI malaria model has a good ability to ...