An Integrated Modeling Environment to Study the Coevolution of Networks, Individual Behavior, and Epidemics
Barrett, Chris, Bisset, Keith, Leidig, Jonathan, Marathe, Achla, Marathe, Madhav, AI Magazine
Contagions (or diffusion) over complex networks are pervasive in social and physical sciences. Three recent global-scale contagions that have received attention in the media as well as academic circles are (1) current and past financial contagions (Desai 2003), (1) (2) failure of the coupled infrastructure system caused by power-grid failure, for example, the Northeast blackout of 2003, (2) and (3) potential pandemics caused by influenzalike illness (Halloran et al. 2008; Germann et al. 1983). Individuals, institutions, and governments could not prevent the Northeast blackout. However, they are aggressively developing interventions to control the current financial contagion and responding to reduce the economic burden and human suffering of the current HIN1 outbreak. Developing high-resolution computational models to reason about these systems is complicated and scientifically challenging for at least three reasons. First, these systems are extremely large (for example, pandemic planning at the scale of the United States, requiring models with 300 million agents). Second, the contagion, the underlying interaction network (consisting of both human and technical elements), the public policies, and the individual behaviors coevolve. This makes it nearly impossible to apply standard model-reduction techniques that have been successfully used to study physical systems. Finally, in practical situations, multiple contagion processes simultaneously coevolve.
Here we describe Simdemics, an interaction-based multiagent approach to study diffusion processes in very large sociotechnical networks. Simdemics is an example of a disaggregated network-based modeling approach in which interactions between every pair of individuals connected in the social contact network are represented. It uses a realistic, synthetic representation of the underlying social contact network. It is based on the idea that a better understanding of the characteristics of the underlying network and individual behavioral adaptation can give better insights into contagion dynamics and response strategies. It should be noted that Simdemics by itself does not prescribe a specific level of quality for the social contact networks. The necessary quality (in terms of accuracy, resolution, and fidelity) of the networks is determined by the questions that we aim to address.
Simdemics can be used to study a much larger class of diffusion processes. These include epidemic processes in ecologies; the spread of certain noninfectious diseases such as obesity and smoking; the spread of fads, conventions, norms, and information in social systems; the spread of worms and malware in communication networks (Channakeshava, et al. 2009). Here, we will confine our discussion to the spread of infectious diseases in human populations. Besides their obvious societal importance, epidemics serve as an excellent example of diffusion processes over interaction networks. Within the infectious disease context, Simdemics details the demographic and geographic distributions of contagion spread. It also provides decision makers with information about the consequences of an outbreak, the resulting demand for health services, and the feasibility and effectiveness of various response options. A unique feature of Simdemics is the size and scale of social and ecological systems that can be analyzed through its use. Planning and responding to the threat of pandemics presents an important societal and public health challenge. Public health authorities around the world are far more prepared to respond to pandemic threats now than they have ever been in the past. However, a number of modern trends continue to make this a vexing problem. These include (1) a larger global population and increased urbanization leading to a higher density of individuals within cities; (2) higher levels of long-distance travel, including international travel; and (3) increased numbers of elderly individuals and individuals with chronic medical conditions. …