Predicting Bat Coronavirus Positivity

Our project pooled data from published studies looking at the prevalence of coronavirus in samples obtained from various bat species. Subsequently, we merged this data with datasets containing information on biological traits, species distribution, foraging information, as well as geographical, ecological, human population and land use attributes. Using this merged dataset, we used factors found to be associated with coronavirus prevalence on univariate analysis to fit a Poisson regression model and a generalized boosting classifier model. These two models were used to predict the prevalence of coronavirus positivity among bats with unknown coronavirus prevalence rates. Finally, applying our model to Rhinolophus bat, we found five Rhinolophus bat species were predicted to have a high prevalence of coronavirus positivity.

Our study took pioneering steps to look at predictive factors for bat coronavirus positivity. Given that the current COVID-19 pandemic is postulated to be related to zoonotic transmission of a bat coronavirus, it is important to understand the factors that make it more likely for bats to have high coronavirus positivity. This may allow us to predict where the next outbreak may strike.