"A Model of Biological Attacks on a Realistic Population "
Assessing the impacts of biological attacks on a heterogeneous city population with diverse demographics in enough detail and fidelity to enable effective response is important from intelligence and planning perspectives. The complexity of the spread and impacts of weaponized disease outbreaks -- particularly contagious ones -- is compounded by the existence of naturally-occuring diseases which may have similar symptoms to those of weaponized diseases. Moreover, the outbreaks are modulated by physical (e.g., road networks and urban geography), social, health, communication, economical, institutional and govermental infrastructures. These infrastructures often are dynamic networks in form. These networks -- particularly social and health ones -- need to be considered in order to systematically reason about the nature and impacts of outbreaks, the potential of media, prophylaxis and vaccination campaigns, and the relative value of various early warning devices. Conventional SIR models in epidemiology do not address this and only operate on the homogeneous sample population level. Multi-agent models provide an effective and ethical system for reasoning about biological attacks on a city. Our model, BioWar, combines state-of-the-art computational models of social networks, communication media, and disease transmission with demographically and spatially resolved agent models, urban spatial models, weather models, and a diagnostic error model to produce a single integrated model of the impact of a biological attack against the background of naturally-occuring diseases on a city. Unlike traditional models that look at hypothetical cities, BioWar is configured to represent real cities by loading census demographics data, social network data, school district boundaries, business location and type data, healthcare infrastructure data, and other publicly available information. Moreover, rather than just providing information on the number of infections, BioWar models the agents as they go about their lives – both the healthy and the infected (each agent has its own spatial -- longitude and latitude -- coordinates). This enables the analyst to observe the repercussions of various attacks and containment policies on factors such as absenteeism, medical web hits, medical phone calls, insurance claims, over-the-counter pharmacy purchases, and hospital visit rates, among others, in addition to epidemiological factors such as infection rate, prevalence, incidence, and death rate. The values of these factors are used in syndromic and behavioral surveillance to evaluate early detection algorithms. BioWar was implemented in multithreaded C++ and achieved fast runtime both on a personal computer and on supercomputer nodes. We will present the validation results of BioWar for anthrax and smallpox outbreaks based on empirical data and comparison against population-based epidemiological models. The results of smallpox and anthrax attack simulations using BioWar on select US cities will be described and analyzed. The results show that BioWar produces higher fidelity and high granularity predictions than what conventional SIR models can attain and produces nonlinear dynamic emergent behaviors and patterns similar to what happen in the real world. Comparisons with other computational models will also be presented. BioWar is thus useful for preparedness training, intelligence planning, response analysis, detection algorithms evaluation, stakeholder communication, and public policy analysis.