How to find environmental risk factors of zoonotic infectious disease quickly

Abstract

Analysis on zoonotic infectious diseases is an important issue in emergency management as it significantly supports governmental and medical decision making when a zoonotic infectious disease outbreaks. To effectively prevent and control the diseases, it is necessary to explore the pathogenesis and identify correlative influence factors. However, limited by natural conditions and physical measurements, we can hardly obtain complete observations to precisely catch on to the actual pathogenesis of zoonotic infectious diseases. A feasible solution for analysis on the diseases is to measure correlations between environmental factors and incidences of the diseases, and then extract the pivotal factors. Many existing studies have provided qualitative analysis on zoonotic infectious disease. In this paper we consider a quantified method using regression models to measure effects that derive from environmental factors. Significant factors are extracted through a multiple backward stepwise logistic regression and compose a set of explanatory variables, which is exploited in the regression of the incidence of zoonotic infectious diseases. Furthermore, considering the variance among different areas and complex interactions between neighboring areas, we incorporate unobserved individual heterogeneity and neighborhood-based spatial effects into the regression model. Therefore, the model is updated with spatial structures. Several different estimators are involved to provide unbiased estimations for models without spatial structures and models with spatial structures. Then comparisons between different models are illustrated. The result shows that our quantified models are valid and the regression model performs better with individual heterogeneity and spatial effects allowed for.