Collegiate Times

A team of researchers at Virginia Tech has created a software program that has been used to predict how diseases are spread by developing a digital simulation of the U.S. population.

Researchers, led by Chris Barrett, director of Tech's Network Dynamics and Simulation Science Laboratory, have created EpiSimdemics, a software program that mimics the movements of about 300 million Americans. The group used data from the 2000 census and other public demographic data to create the program.

Barrett said the software is trying to answer the question, "What interacts with what? In terms of an airborne disease like influenza or cold, the question is where are people co-located."

The software has been used to simulate how pandemic flu might spread throughout a population and how interventions function to change it.

Each synthetic individual in the software has about 68 base variables that are taken from the census. Depending on the question that is posed to the researchers, variables are added or taken away. Every home and business address in the United States is represented in their model.

"It's an incredibly detailed, understandable representation of all of the complexity of life with all of the characteristics of life with very realistic individuals in it so we can understand things. Like what happens when you close the schools, what happens if you close these roads because of an exposure to a biological attack," Barrett said.

The research has been used to analyze how medicine is distributed and how that would affect the spread of the disease. The simulations looked at distributing medication to only the sick and letting citizens buy medication on their own. They then took the data to see which method of distribution decreased an individual's chance of dying.

The research has tracked synthetic social networks to monitor how communicable diseases are spread among people. Data took into account whether the individual drove to work or used mass transit and whether or not schools were closed -- both factors that contribute to how a disease is spread. Primarily, the researchers have been working with pandemic flu.

"There's a lot of fear that the current strains of avian flu might spread toward a worldwide infection," said Bryan Lewis, a Tech graduate student and an associate researcher in public health. "Our social network is so realistic, we say we have a better estimate of what it would look like in real life."

So far, avian flu has only been spread from birds to humans and has yet to be transmitted from human to human.

While their representation of the population is detailed, Barrett said their research is not invasive and protects privacy.

"This stuff is in a form that protects individual privacy. So privacy (and) anonymity of sources are protected," Barrett said. "It is generally speaking, simply it's a very realistic structural representation of the population at an individual level, but it isn't the real population. There are things that are very accurate, but on the individual level I can't predict what you particularly are going to do with this simulation."

The Department of Health and Human Services, Department of Energy, Department of Defense, Department of Transportation, National Institutes of Health, and Centers for Disease Control have all worked with Tech's Network Dynamics and Simulation Science Laboratory.

One of the challenges for software developer Keith Bisset is to ensure that results presented to clients are not biased.

"The difference between saying an intervention that you did saved 80 percent, but it killed 20 percent, it's the same set of data, but it changes the way people think about it," Bisset said.

Outside of the realm of public health, the research has also been applied to energy markets and power grids, wireless networks and traffic flow.

One of their earliest research programs was to study traffic flow for the Department of Transportation. This was used by transportation officials to determine if they should build roads in certain areas to help decrease traffic in a specific area. Lewis used the example of whether or not adding another lane to 460 would decrease commute time.  

Both Bisset and Lewis said they hope their software may be used to simulate how different diseases, such as malaria, are spread throughout developing countries.

"There's lots and lots of big questions that we'd love to be involved in, including major world health issues," Barrett said. "There are lots of big questions in the world and lots of them that we could maybe help with."