Networking in the Feedlot

Computer scientists refine software to predict the spread of livestock diseases



Remember those shocking media images from Britain of cattle herds being culled during the 2001 outbreak of foot-and-mouth disease? Farmers and policy-makers looking to better contain and treat such infectious livestock diseases ― and avoid unnecessary animal deaths ― might look to recent studies by Joel Francis, a newly minted master’s graduate in Computing and Information Science (CIS).

In research completed this spring, he aims to refine a computer model simulating how infectious diseases from foot-and-mouth to bird flu will spread through livestock populations.

That particular software simulation was developed by Guelph computer scientists led by CIS Prof. Deb Stacey, Francis’s grad supervisor and chair of the department, along with other researchers in Canada and the United States. The North American Animal Disease Spread Model (NAADSM) is now used by epidemiologists and policy-makers worldwide, including the U.S. Department of Agriculture and the Canadian Food Inspection Agency.

“It’s used all over the world,” says Stacey, who helped build the simulator with collaborators at Colorado State University using federal funding provided after the 2001 British outbreak. Explaining that computer simulations can help to study how a livestock disease outbreak might spread, she says: “This is a tool to help you think about the problem.”

Sharpening that tool was the goal for Francis. He started working with Stacey in 2008 after completing a degree in applied computing at the University of Guelph-Humber.

“We’re using computing science to simulate and build models and study the spread of an outbreak,” says Francis.

To do that, he ran simulations on the SHARCNET computing installation that connects clusters of computers at Guelph and 16 other universities, colleges and research institutions in Ontario. The Shared Hierarchical Academic and Research Computing Network is among the most powerful supercomputing systems in the world.

That kind of computing power is necessary for looking at patterns of connections linking farms and other players moving animals around in the livestock industry. Those connections are not random but may show “scale-free” properties.

Scale-free networks have nodes or hubs with more connections than others. Just think of the who-knows-whom circles connecting users of social networks such as Facebook. “Some things are highly connected, but most things are not,” says Stacey. “Everybody is connected to Google, but Google is connected to a few people.”

To understand how a network operates and the effect of policies such as livestock vaccination or culling, she says: “It’s very important to know what the underlying connection is because it makes a big difference.”

In the Canadian countryside, the livestock industry consists of larger shipping or supplier hubs that work with many individual farms. Francis found that the scale-free structure affects the performance of various disease containment strategies. Surprisingly, his sensitivity analysis showed herd vaccination was less important than other variables, such as population density, shipping distance or hub size.

“You need to know who’s connected to whom,” says Francis, adding that his work might help modellers experiment with networks in developing nations lacking precise information about farming infrastructure.

Stacey plans to continue working on the animal disease model with research associate Neil Harvey, adjunct professor and PhD grad Greg Klotz, and other current and former grad students. Late last year she received new funding from the Poultry Industry Council to look at disease spread among flocks.

Francis will continue that research while he pursues work in IT or wireless communication. Originally from Qatar, he lives in Toronto.