An Agent-based Modeling Approach to Estimating Combined Prevention Effects
Learn more about the project “Agent-based Modeling of Firearm Mortality” presented by Jason Goldstick at the Firearm Safety Among Children and Teens (FACTS) symposium, September 23, 2020.
In this project, a statistical modeling strategy will be used to simulate the hypothetical effects of both policy and public health interventions on youth firearm mortality in an artificial city demographically and spatially mirroring mid-size rust belt cities. This will be accomplished by developing an agent-based model (ABM) or simulation model of firearm violence to better understand the underlying causes and solutions to interpersonal firearm violence among adolescents.
Agent-based modeling (ABM) is a powerful tool that can be used to simulate complex systems and to test hypotheses that may be intractable or infeasible to test in practice. In particular, ABM is particularly well-suited to i) model individuals’ movement and interaction within a spatial landscape; and ii) simulate and evaluate counterfactual conditions, such as policy-based interventions. Such models have played a prominent role in many health applications, including studies of obesity, causes of disease, physical activity, pedestrian flow, walking behavior, land use management, and property crime (cite). Currently, there is an increasing call for their application to public health research at large and, in particular, social epidemiology. Yet, ABMs have not been applied to the prospective evaluation of hypothetical firearm policy interventions.
This project proposes the development of an ABM of violence in an artificial city mirroring the structure of Flint, Michigan. The specifics aims are:
Aim 1: Develop an ABM of firearm violence in an artificial city mirroring Flint. We will develop the ABM described above and calibrate the unknown parameters to reproduce the spatio-temporal distribution of firearm violence crime observed in police-reported crime incident data observed in Flint, Michigan.
Aim 2: Simulate CAP/safe storage and minimum purchase age restrictions on pediatric firearm violence using the model. Based on the calibrated model, simulate both policies simultaneously to determine synergistic (or antagonistic) interactions between the two in terms of their impacts on pediatric firearm violence; determine how the effect on firearm injury compares with effects that are possible using one or the other policy and use this information to form the basis of a cost analysis.
This work is significant because it will provide a basis for prospectively determining the potential effect, and cost effectiveness, of counterfactual conditions on the second leading causing of death among youth in the US. The innovation of this work is its focus on previously unexplored simulation-based approaches to the study of firearm policy among youth; this innovation is important because it can form the basis for other analogous studies into the hypothetical impact of other policy- and non-policy-based interventions on youth firearm injury.
Michigan State University
University of Washington