Using GIS, Simulation of Flu-Like Illnesses Like SARS Draws a Nuanced Picture of Public Health Threats

BUFFALO, N.Y. -- A new, computational method for simulating the spread of flu-like illnesses like SARS (severe acute respiratory syndrome) that is being developed by a University at Buffalo geographer may provide policymakers and analysts with new ammunition for studying and predicting the pattern of public-health threats in urban communities.

The research, described in a paper in press at Environment and Planning B, uses the tools of geographic information science (GIS) and object-oriented computing to create a realistic picture of how an infectious, flu-like illness would spread throughout Buffalo.

"This type of model allows us to foresee in a more nuanced way what type of risk a community may face," said Ling Bian, Ph.D., associate professor of geography in UB's College of Arts and Sciences and author of the paper.

The model she is developing differs from some more conventional epidemiological models by taking into account characteristics and behaviors of individuals, the relationships between them and with their environments and how those interactions change over time and space.

Since the model more explicitly simulates differences in human interaction at different times and at different locations, it could play a role in helping develop policies to contain more effectively or reduce public health threats, she said.

According to Bian, since the spread of an infectious disease throughout a community is a spatial process, GIS is suited uniquely to demonstrate it.

For example, a key contribution of her model is its use of GIS to take into account both the daytime and nighttime locations of individuals. Many conventional simulations include just the nighttime, or home location, of individuals.

This use of multiple points of contact for an individual allows for a realistic representation of how an infection might spread through a community based on the numbers of people with whom an individual regularly interacts both at home and at work.

Bian's simulation also employs object-oriented computing, which allows the model to focus on individuals in a population rather than the population as a whole.

Therefore, she explained, it is capable of representing the variability of the spread of an illness throughout a community, based on specific attributes of individuals, such as their age and susceptibility to infection, daily travel routine and people with whom they come into contact.

In addition, she said, the pattern of an infection may spread very differently in different communities, based on specific demographic features.

For example, Bian explained, in a more rural suburb of Buffalo, building lots are large so neighbors may interact less than they do in denser suburbs, where lots are small and children from different families play together in a small area.

Bian's simulation assumes that most people have a moderate number of social connections, while other models often assume a "global mix" -- that everyone in the community has contact with one another.

"Our model simulates social connections," said Bian. "Everyone has a network of human contacts. On a daily basis, you travel between home and the workplace and in each place you have contact only with a limited number of people, even though you may work in a large institution."

However, she noted, living arrangements also differ within communities, such as retirement homes where there may be fewer outside contacts than a family of four, and college dormitories, where individuals may have more outside contacts.

Bian's model can be adapted to represent these differences and their influence on how flu-like illnesses travel.

The research was funded by the National Institute for Environmental Health Science of the National Institutes of Health.