In search for the hotspots of Disease X: A biogeographic approach to mapping the predictive risk of WHO's blueprint priority diseases

Abstract

Background: Current trends of emerging infectious disease outbreaks (EIDs) forecast impending global epidemiological crises. Human-driven environmental changes, including climate change along with overpopulation and global travel, have been contributing to EIDs outbreaks in many developing countries. The subject has attracted increasing attention with the recent Ebola and Zika epidemics, which highlights the potential threats to human and animal health, social stability, and global trade and economy. The blueprint priority diseases (BPDs) is a list established by the World Health Organization of ten zoonotic diseases, which are in urgent need of research. We proposed mapping the predictive risk of the BPDs using spatial Bayesian models and species distribution modelling of the outbreaks following the year 2000. The aim is to provide a global perspective, measure predictive risks, and evaluate the use of biogeography on predicting diseases outbreaks. We also proposed disease biogeography as a tool for identifying the potential hotspots for Disease X listed in the BPDs. Methods and materials: Data of the observed outbreaks (2000–2018) were obtained from promed mail and WHO archives. Bioclimatic covariates and altitude data were extracted from ‘worldclim’ (R package dismo) and the 2017 land cover data (MODIS satellite imagery). We constructed species distribution models including bioclim, maxent and Bayesian models with absence data generated to map the predictive risk of future outbreaks. Results: Most of the predicted geographic risk extent estimated from the observed data of MERS, Marburg Virus Disease and Rift Valley fever were found to occur in arid and across the Middle East and Eastern regions of Africa. The predictive extent of Lassa fever and Ebola Virus Disease consisted of regions in the Western and Central Africa, predominated by humid rainforests. We found a significant correlation between the disease extent and the distribution of confirmed and suspected biological reservoirs and also with deforestation. The AUC of the generated models was maintained over 0.9 (average - 0.978). Conclusion: Biogeography is a robust tool in forecasting hotspots of EIDs outbreaks. We will also complete the analysis by aggregating the common risk factors of the predicted EIDs to characterize hotspots for an unknown Disease X.