Risk Factors for African Swine Fever in Wild Boar in Russia: Application of Regression for Classification Algorithms

Abstract

The population density of susceptible animals, including domestic pigs and wild boar, is a major risk factor for the emergence of African Swine Fever outbreaks. The ASF foci in wild boar in Russia is sustained by the presence of the virus in the environment, which is primarily determined by the number of infected carcasses found. This study investigates the risk factors related to the occurrence of ASF virus among wild boar, identified through passive monitoring and depopulation control measures, by employing generalized logistic regression models and random forest analysis. The random forest regression outperformed logistic regression coefficient of determination (R2 = 0.98 and R2 = 0.88) according to the statistical modeling of ASF using different regression types. When comparing regression models, the results showed that wild boar population density, the number of hunting farms, the presence of infected carcasses, and ASF outbreaks among domestic pigs were the main predictors of epidemic in wild boar. The application of a multiple logistic regression model confirmed the significance of the identified risk factors, determining of the probability of ASF outbreaks among wild boar. Given the prolonged affected area across most regions of Russia, the random forest model proved to be the most effective and interpretable based on quality indicator assessments. By highlighting the important role of geographical conditions, identifying these risk factors enhances our understanding of ASF dynamics in specific regions and offers valuable information for decision-makers in developing targeted control strategies against this disease.