Agent-based Modeling and Infectious Disease Agents

Agent-based Modeling and Infectious Disease Agents

  • 29 September 2013

Quantitative and analytic methods represent tools that can be used to understand a broad range of processes that depend on interactions between individuals or groups of people, for instance from economic change to social disorder. In this paper, the use and advances based on these tools for understanding the spread of disease is discussed, focusing on research conducted as part of the Models of Infectious Disease Agents Study (MIDAS). The rising threats of avian influenza (bird flu), SARS-like diseases such as the currently circulating MERS-CoV, and concerns about a biodefense emergency convey the need to prepare for, detect, and confront rapidly spreading, infectious diseases.

With the advent of cheap and powerful computers, a new set of methods called agent-based models (AMBs) have become commonplace and have emerged as powerful alternatives to traditional modeling approached.  Two compelling traits of ABMs are (1) the use of flexible population-mixing rules of behaviors, and (2) the ability to describe population movements and contacts at a close and personal level of detail. Unlike traditional modeling approaches, adding new population groups and agent behaviors to epidemiologic ABMs is straightforward and has resulted in accurate portrayal of heterogeneous (mixed) populations. Consequently, ABMs have made a number of key contributions for evaluating potential responses to emerging infections, estimating key epidemiologic properties of emerging infections, and characterizing features of infectious diseases that either make them likely to evolve toward successful human-to-human transmission or that affect the feasibility of control measures. New technologies, such as social media (Twitter), can provide data feeds to the models to detect outbreaks and update the simulations.

Because agents in the epidemiologic ABMs are people (although they may also be vectors of disease, such as mosquitos), detailed population data are often not available because of confidentiality concerns. So, methods to generate realistic “synthetic populations” were developed which imbue agents with demographic characteristics, as well as organize the agents into social networks such as households, schools, and workplaces.  Depending on the level of detail in the data sources used to build the synthetic populations, geographically-specific and time-specific differences can be captured for very small areas. These synthetic populations can be overlaid with behaviors—such as international airline travel or domestic commuting, participation in public health initiatives like vaccination, or self-isolation in response to outbreaks—to build models that can be used to test impact of various mitigation and control responses.

Abu Dhabi is an important nexus of travel in and out of the region and around the world. With the influx and outflow of individuals, any one of whom may spread transmissible disease, the utility of ABMs to help understand the spread and control of infectious diseases should be part of the tool armamentarium. This paper examines the experience using the ABMs to understand recent epidemics and current research to develop synthetic populations. Changes to the synthetic population that mirror large social events when people congregate and disperse, such as the Hajj, are also being modeled.

The principles and research described are based, in part, on research conducted by RTI International, in particular by Philip Cooley and William Wheaton.

Share

LECTURER

Sunday 29 September 2013

-

Sunday 29 September 2013

-