Computer Simulates Bird Flu Spreading Across U.S.

A new computer model reveals how a pandemic caused by the avian flu virus might spread quickly across the United States, and also what methods would best thwart the disease.

Researchers assumed a starting point of 10 highly infectious influenza cases in Los Angeles, then let the model take it from there.

The virus spread quickly, peaking in just 90 days with 100 or more infections per 1,000 residents in just about every corner of the country.

• Click here to watch the video (QuickTime .mov format).

The simulation is an attempt to map out what might happen with a very uncertain bug. The specific strain ravaging the world's bird populations, H5N1, does not yet easily pass between humans.

If it does mutate and develop that ability, however, human deaths could mount quickly. Vaccines developed for existing strains would likely not be effective.

• Click here for the Bird Flu Center.

U.S. Health and Human Services Secretary Mike Leavitt has said the country is not prepared for such a scenario.

Lessons learned

The virtual pandemic suggests that advance stockpiling in large quantities of a modestly effective vaccine would be preferable to waiting to see exactly what strain emerges.

Quarantines, school closures and travel restrictions alone wouldn't thwart the spread of human-to-human avian flu, but such measures could buy time while vaccines production was ramped up and tailored to the specific flu strain.

In the simulation, long-range travel by humans was cut to 10 percent of normal, based on travel advisories that would presumably be instituted.

"Based on our results, combinations of mitigation strategies such as stockpiling vaccines or antiviral agents, along with social distancing measures, could be particularly effective in slowing pandemic flu spread in the U.S.," said Ira Longini, a biostatistician with the Fred Hutchinson Cancer Research Center and the University of Washington.

The variables

The model uses data from the 2000 U.S. Census posits 281 million people at work, play, school and home, along with Department of Transportation travel data that incorporates rapid spread from one city to another by air travel.

The computer model employs probabilities that an infected person would cross paths with others at home or, with lower probability, elsewhere.

"So we are only computing the probability of any person becoming infected on any given day, and a roll of the dice is needed to decide whether they are infected or not," said Timothy Germann of the Los Alamos National Laboratory.

The computer also considered one vexing aspect to the flu: About 33 percent of those infected don't develop symptoms and can unknowingly transmit the disease.

[The simulation was conducted using the lab's Scalable Parallel Short-range Molecular dynamics (SPaSM) platform, originally developed to simulate nuclear explosions. The hardware used was Pink, a Linux-based supercomputer cluster at the lab, which is made up of 1,024 nodes, each of which has two 2.4 Ghz Intel Xeon processors and two gigabytes of memory.]

Copyright © 2006 Imaginova Corp. All Rights Reserved. This material may not be published, broadcast, rewritten or redistributed.