From mathematical models to public policy
Policymakers are ultimately concerned with decision-making and they need access to the best available information. In the public health arena, the Models of Infectious Disease Agent Study (MIDAS) group has to connect with a diverse group of decision-makers spanning national and state levels, local health departments, and researchers from different disciplines. To be useful, computer simulations have to be put in place that meet the needs of this diverse group of stakeholders. Everyone has to be speaking a common language and aware of what each group brings to the table.
The conversations are far-reaching. The NDSSL at VBI has been advising a National Institutes of Health steering committee for developing modeling methods for contagious diseases. It has also been giving advice on pandemic influenza planning to the U.S. Department of Health and Human Services, the Department of Defense (DoD) and the White House. Two large studies to support pandemic planning for military preparedness were recently completed for the DoD’s Defense Threat Reduction Agency using a prototype system known as the Comprehensive National Incident Management System (CNIMS). This system integrates surveillance, simulation-assisted hypothesis testing, and decision support for use in situational awareness and planning of complex systems.
“The computational tools we provide originate from statistics, advanced algebra, math, and complexity science. What we deliver at the other end is a resource that needs to be practicable and directly relevant for policy- and decision-makers. This is a service-driven process and, as computational scientists, we have to work with all of our partners from day one to deliver the best possible tools that meet their needs,” says Chris Barrett, director of the NDSSL. “We put a lot of effort into dialog with federal agencies, the government, public health experts in the front line, and other end-users to develop our modeling and simulation tools,” he adds. “Like any good dialog, the modeling process is never complete and you continually have to make refinements. It’s not so much about how things should work in the perfect world but more about understanding how things work in the real world. To make it work, you have to get the difficult questions and they need to come directly from the policy- and decision-makers as well as public health professionals.”
No model is perfect. “Models of pandemic influenza are put together from the best available data and the best available computational resources,” says Eubank. “For pandemic influenza, we have been able to build up realistic pictures of how social mixing patterns change under different interventions.
“Although they contribute in a big way to understanding how influenza would spread through cities and populations, they are not the only resource,” Eubank says. “Scientists, field workers, and other experts around the globe are working on many different fronts to tackle the influenza virus and that will be critical for success.”
Phase 6 readiness
Advances in computational resources and methods mean that researchers are reaching a point where they will be able to perform these computer simulations on a national scale. “Until now, the amount of detail that we’ve needed has meant working with one major city at a time. However, we have continued to refine our synthetic populations and computational algorithms and we are reaching the point where such studies are feasible for the whole population of the United States,” Eubank says. “Big clusters of computers that are organized into grid-like systems make countrywide studies possible. Tools such as the Comprehensive National Incident Management System will move this capability out of research labs and onto an analyst’s desktop.” Experts believe that the world is closer now to another flu pandemic than at any time since 1968, when the most recent pandemic took place. For now, the World Health Organization says that the world is on phase 3 alert. This means that a new influenza virus subtype, the avian H5N1 bird flu virus, is causing disease in people but is not yet spreading quickly among humans. If efficient and sustained human-to-human transmission were to materialize, a phase 6 alert would be issued. Computer modeling and simulations like those performed by VBI allow researchers to enact phase 6 scenarios today. The hope is that by identifying the best possible combination of interventions, lives can be saved when the next influenza pandemic strikes.
Image shows simulated social contacts for an individual above the Chicago skyline. Simulation courtesy of MIDAS and San Diego Supercomputer Center.