A team of USC engineers placed in the top 2 percent among 466 teams in a national challenge, earning an invitation to meet with government experts in disease prediction at the Defense Advanced Research Project Agency conference center in Virginia.
USC Viterbi School of Engineering Ph.D. students Charalampos Chelmis and Anand Panangadan and computer science Ph.D. candidate Ajitesh Srivastava were recognized for producing a prediction model that uses information sharing to accurately deliver a six-month forecast of the spread of the Chikungunya virus in 55 different countries and territories in North, Central and South America and the Caribbean.
“It’s always a big boost to see your research ideas being applied to the real world and used in different practices,” Chelmis said.
Thousands of confirmed cases
As of May, there have been more than 33,000 confirmed cases and close to 1.4 million suspected cases of the Chikungunya virus in the Western Hemisphere since its first appearance in December 2013, according to DARPA. Spread by mosquitos, Chikungunya is rarely fatal but can cause joint and muscle pain, fever, nausea, fatigue and rash.
DARPA is an agency of the U.S. Department of Defense responsible for developing emerging technologies for the military. It began the CHICKV Challenge in 2014 to create models that could accurately predict the spread of Chikungunya.
Panangadan said the USC Viterbi team used concepts of information sharing on the Internet, applying them to predicting the spread of viruses.
Srivastava, who won an award for best presentation in the competition, compares the prediction of such outbreaks to a shared YouTube video spreading rapidly across the Web.
“Information sharing on the Internet is based off of information sharing on epidemics, so we took that and applied it to predicting outbreaks,” Srivastava said. “It’s a similar method to the process of assessing how a meme or a video goes viral.”
Chelmis, Srivastava and Panangadan, along with the other winners, shared their work with members of the Center of Disease Control and the U.S. Department of Health to raise awareness of epidemic prediction and to foster potential future collaborations between the teams and the experts.
“Our goal was to make our model interactive so a person who doesn’t have a background in math or epidemiology can use it and understand its findings,” Srivastava said. “When you wake up and there’s a zombie apocalypse, Rick Grimes [The Walking Dead] won’t be able to save you, but our project can help people learn how it’s spreading.”