Johns Hopkins APL Team Visits Washington for Dengue Prediction Challenge Win

In September, research by Johns Hopkins APL staff members (left to right) Linda Moniz, Anna Buczak, and Erhan Guven (and Ben Baugher and Thomas Bagley, not pictured) on how to determine the potential spread of dengue fever was recognized as part of the Dengue Prediction Challenge, sponsored by the the White House Office of Science and Technology Policy, the Centers for Disease Control and Prevention, the Department of Defense, and the National Oceanic and Atmospheric Administration. Courtesy of JHU/APL; image of dengue virus used with permission of Howard Hughes Medical Institute

Research from the Johns Hopkins University Applied Physics Laboratory (APL) in Laurel, Md., into better methods of predicting outbreaks of the mosquito-borne dengue virus was selected for presentation in September at the Eisenhower Executive Office Building - part of the White House complex. The APL team had developed new options for forecasting the spread of dengue fever, which affects up to an estimated 390 million people annually worldwide.

The presentation was the culmination of the Dengue Prediction Challenge, organized by the White House Office of Science and Technology Policy, the Centers for Disease Control and Prevention (CDC), the Department of Defense, and the National Oceanic and Atmospheric Administration. The challenge was designed to find ways to improve efforts to predict dengue epidemics and potentially improve public health outcomes.

According to the CDC, dengue viruses are on the rise, spreading among humans through mosquito bites. In the United States, recent outbreaks have occurred in Florida, Texas, Hawaii and Puerto Rico. With no vaccine available to prevent dengue fever, the CDC notes one of the best ways to reduce the disease’s impact is to prepare health care providers by forecasting epidemics before they happen.

Earlier this summer, several U.S. government agencies challenged the infectious disease community to design disease forecasting models that could improve dengue epidemic predictions and potentially improve public health outcomes.

Over a dozen universities across the country, as well as several U.S. and European companies, participated in the four-month project. The winning APL team consisted of Anna Buczak, Ben Baugher, Erhan Guven, Thomas Bagley - all from APL’s Asymmetric Operations Sector - and Linda Moniz, of the Lab’s Space Exploration Sector. Bagley, a Duke University computer science student, was a summer intern at APL at the time.

To develop their forecast models, participants were given access to historical dengue surveillance data (dengue case counts and environmental, climatological, demographic, and vegetation data) from San Juan, Puerto Rico, and Iquitos, Peru. The teams had to predict three variables for each location: timing of peak incidence, maximum weekly incidence and total number of cases in a transmission season. When final submissions were evaluated in September by representatives from numerous government agencies, the APL team’s predictions were selected as winners for two of the Iquitos, Peru, variables: maximum weekly incidence and the total number of cases in a transmission season.

“The method we developed for predictions for both locations was the same,” says Buczak, who led the APL project team. “Iquitos was much more challenging to predict than San Juan because the time series were much shorter, and the data were substantially noisier. We developed an ensemble method, in which each ensemble was made from 300 best models.”

The APL team shared their viewpoints at the meeting, which included representatives from the National Science and Technology Council’s Interagency Pandemic Prediction and Forecasting Science and Technology Working Group. They also offered potential next steps to strengthen infectious disease forecasting.

The government sponsors plan to develop a manuscript to publish the evaluation results. All submitted forecasts will be included, and all team leaders will be invited as authors.

Source: Johns Hopkins University Applied Physics Laboratory