Estimating Cyclic and Geospatial Effects of Alcohol Usage in a Social Network Directed Graph Model

Abstract

Alcohol use and abuse contributes to both acute and chronic negative health outcomes and represents a major source of mortality and morbidity in the world as a whole (Ezzati et al., 2002) and in the developed world, such as the United States, where alcohol consumption is one of the primary risk factors for the burden of disease. It ranks as a leading risk factor after tobacco for all disease and premature mortality in the United States (Rehm et al., 2003; Said and Wegman, 2007). In a companion piece to this paper, Wegman and Said (2009) outline a graph-theoretic agent based simulation tool that accommodates the temporal and geospatial dimensions of acute outcomes. In order to complement that modeling paper, this paper focuses on methods to exploit temporal and spatial data with the idea of calibrating the model to include these effects. The model developed inWegman and Said (2009) incorporates a social network component. The overall goal of the model is not just to simulate known data, but to provide a policy tool that would allow decision makers to examine the feasibility of alcohol-related interventions. Such interventions are designed to reduce one or more acute outcomes such as assault, murder, suicide, sexual assault, domestic violence, child abuse, and DWI-related injuries and deaths. By adjusting conditional probabilities, the effect of interventions can be explored without actually introducing major societal policy actions.