View Single Post
Old 06-02-2015, 03:24 PM
campbell's Avatar
campbell campbell is offline
Mary Pat Campbell
Join Date: Nov 2003
Location: NY
Studying for duolingo and coursera
Favorite beer: Murphy's Irish Stout
Posts: 89,609
Blog Entries: 6


How Computer Modelers Took On the Ebola Outbreak

Did real-time epidemic modeling save lives in West Africa?

Here’s what we know for sure. On the afternoon of Friday, 3 October 2014, Pyrros A. Telionis got a telephone call from the U.S. government’s Defense Threat Reduction Agency (DTRA). He was sitting in a featureless cubicle in Blacksburg, Va., that would have made the cartoon office drudge Dilbert feel right at home. The voice on the phone was brisk and professional. And highly specific. Could Telionis provide, by 8 o’clock Monday morning, a list of the best places to build Ebola treatment centers in Liberia’s six southernmost counties?

There may or may not have been a U.S. Air Force cargo plane on a tarmac somewhere, loaded with construction materials, awaiting Telionis’s list of locations. IEEE Spectrum couldn’t confirm that part of the story.

At first glance Telionis, a Ph.D. student in computational biology at Virginia Tech, might not seem the obvious person to advise the Department of Defense, which was preparing to send about 3,000 people to West Africa to help combat the worst outbreak of Ebola virus the world has ever seen. The military personnel planned to build Ebola treatment centers, mobile labs, and a hospital for infected health care workers.

Telionis did, however, bring hidden strengths to the Defense Department’s urgent assignment. Behind him was the expertise and computing power of the Network Dynamics and Simulation Science Laboratory, which is part of the 250-person Virginia Bioinformatics Institute. Split between Virginia Tech’s Blacksburg campus in rural southwest Virginia and an office building near Washington, D.C., the lab has done epidemic modeling for the Pentagon for nine years. Just that week, Telionis had run an experimental program he’d written to determine the best locations for Ebola treatment centers in a few Liberian counties. DTRA was impressed by the exercise, and now the agency wanted him to do it for real.

First, Telionis had a decision to make. It was the last day to drop classes at Virginia Tech without penalty. He abandoned Advanced Methods of Regression Analysis and spent the next 64 hours working for DTRA.

Did the Virginia Tech models make a difference to the Ebola response? We may never know for sure. The map of the treatment centers that the Defense Department built certainly looks a lot like the one that Telionis and Schlitt turned in. Beyond that, it’s impossible to say, because Defense Department officials won’t comment.

The experts swear their models will work better next time. The Virginia Tech group is now creating an epidemic model that will represent the entire world and its complicated human occupants. Every country will have a “synthetic” population that stands in for the real one. There will be the right number of elementary school students per square mile, the right number of women in workplaces with more than 50 employees, the right number of households with fewer than five people, and myriad other demographic data slices. The model will also attempt to capture, at least roughly, what people do each day—the probability of leaving the home, going to a farm, getting on a bus or an airplane. Researchers don’t have those details for many places and will often have to infer them. For example, in estimating how many people labor in Liberia’s various types of workplaces, the Virginia Tech modelers are using data from Mexico. In other cases, an available database can give analysts a good approximation of unavailable information. Tracking the locations from which cellphone calls are made, for instance, gives a good moment-by-moment view of how people are moving in a region.

These “agent-based” models will give a more nuanced picture of how pathogens affect and sicken a population. “This is the wave of the future,” says Stephen Eubank, deputy director of the Virginia Tech lab. “It’s going to take a concerted effort to gather the data and the expertise. But it’s going to happen.”

And so, too, will another Ebola outbreak.

This article originally appeared in print as “Computer Modelers vs. Ebola.”


LinkedIn Profile

Last edited by campbell; 06-02-2015 at 03:28 PM..
Reply With Quote
Page generated in 0.24019 seconds with 9 queries