October 23, 2014
The workshop, attended by close to 50 international scientists and public health officials, brought together federal and non-federal stakeholders, including researchers from the U.S. Department of Defense, the Centers for Disease Control and Prevention, and the National Institutes of Health, as well as leading experts from top-tier research universities around the world. The goal of the workshop series is to accelerate the development of models that could be used for predicting the occurrence of outbreaks of infectious diseases, such as dengue fever. The day-long session included presentations on current dengue surveillance efforts in Peru and Thailand. Organizers planned the exercises to contribute to the broader objective of strengthening infectious disease prediction and forecasting models to support public health and national security decision-making.
“It was a thrill to participate in this workshop and see so much collective wisdom on infectious disease modeling and dengue in one place at one time,” says Reich, an international expert in statistical modeling of infectious disease data. His analysis of a unique data set spanning 40 years of dengue fever incidence in Thailand recently led to the discovery of the duration of temporary immunity provided by an initial dengue infection, a research breakthrough with particular implications for designing more effective vaccine studies. “It is exciting that the U.S. government is recognizing the importance of infectious disease prediction. Even just asking the seemingly simple questions about what the outputs from such forecasting efforts could be and how might they be used in public health practice is a vitally important conversation.”
Dengue fever, a viral infection transmitted to humans by mosquito bites, often causes severe symptoms including a rash, fever, headache, and muscle and joint pain. Dengue circulates in over 100 countries and is emerging in much of the western hemisphere including places such as Puerto Rico and Florida. As global temperatures warm due to climate change, dengue may become endemic in more parts of the world. Dengue fever is responsible for an estimated 50 million infections and 19,000 deaths worldwide each year. 2.5 billion people worldwide live in regions where dengue is endemic.
Recent large dengue epidemics in India and Thailand resulted in significant morbidity and mortality and put a tremendous strain on healthcare systems and local and national agencies. Increasingly, these outbreaks are seen as having a substantial impact on national and global economies. Despite the potential consequences, creating accurate disease forecasts remains extremely challenging because many different factors impact the ability of dengue to spread. Also, Reich points out, not all case data may be available at a given time due to delays in the reporting process of the surveillance system. For example, it may take weeks or months for cases to appear in an official surveillance record.
“Nate Silver always says that his task of forecasting the outcomes of U.S. elections is easy, and I think he’s right. In comparison, the spread of infectious diseases are much less well understood, we have much less data to work with, and instead of predicting a simple ‘yes or no’ outcome, we are trying to pinpoint exactly how many cases may occur in a given week several months in the future. That’s a tall order, especially given how much something like the weather can impact these disease transmission processes. One of the biggest challenges we face as researchers is accurately representing the uncertainty in our forecasts, which may be substantial, especially far into the future.”
Several federal departments and agencies in the U.S. are developing capabilities to monitor, predict, or forecast dengue outbreaks. A goal of the workshop series is to identify and make available datasets that could strengthen new and ongoing modeling efforts. Mosquito biology, demographics, weather, climate, and travel patterns play an important role in transmission and could be included in the development of prediction and forecasting models. Further, enhanced dengue modeling could provide important insights into the spread of other vector-borne diseases, including chikungunya, another mosquito-borne virus that spread throughout Central and South America for the first time in 2014.
“Compared with other infectious diseases, we have a lot of data and pre-existing knowledge about how dengue circulates in a few particular places in the world,” notes Reich. “While it is still a really complicated and dynamic system, the surveillance systems that track dengue are some of the best and oldest in the world, giving us really rich data to work with.”
After an interim phase designed for data collection, Reich and colleagues will reconvene in the spring of 2015 for a modeling workshop to share data, fit and evaluate the models, and assess both the results and the process.
“I’m excited to see where we can go from here,” says Reich. “This is a situation where letting modern statistical techniques loose on streams of real-time data has the potential to make a tremendous impact on disease surveillance and public health practice around the world.”