Reich seeks to improve infectious disease predictive models

University of Massachusetts Biostatistics faculty Nicholas Reich

Nicholas Reich

December 1, 2016

Courtesy UMass Amherst News Office

Assistant Professor of Biostatistics Nicholas Reich at the University of Massachusetts Amherst School of Public Health and Health Sciences, one of the leading researchers in the nation developing prediction models for infectious diseases such as influenza and dengue fever, recently received grants totaling more than $2 million to create better prediction methods for infectious disease worldwide.

Part of the goal, he says, is “to build a new framework with an eye toward delivering useful results to public health practitioners.” Results should be useful for hospital epidemiologists and others responsible for making public health decisions. Reich’s five-year, $1.9 million grant from NIH’s National Institute of General Medical Sciences grant will support his laboratory in conducting a broad set of investigations to create better predictive modeling for infectious diseases by combining different models.

As he explains, “We are working on implementing what I think is kind of a common-sense approach to prediction. We want to create different models that can give us guesses or predictions about what the coming flu season will look like. Is it going to be a big year or a small year? We are kind of taking a ‘wisdom of the crowds’ approach and averaging all the guesses we get.”

“But, we’re going to try to be clever about how we combine the guesses by looking at which models have historically done better than others at different times of year,” he adds. “Using flu as an example, early in the season there is not much information because there are few cases. So often your best guess at how big the season will be, is just an average of the last handful of years. But once the season gets started, then you start to get a bit more information and models that take the most recent information into account will often perform better.”

In hospitals, a flu outbreak triggers precautions and procedures such as having more gloves, masks and gowns available, adding special decontamination procedures and reducing visitor access to units, for example. These cost time and money, Reich says, and hospitals want to be as effective and efficient as possible in their preparations and response.

His lab works closely with officials at the Centers for Disease Control and Prevention (CDC) and is participating for the second consecutive year in their FluSight challenge, where different teams try to predict seasonal influenza in the United States. Last year, Reich’s postdoctoral research associate Evan Ray developed a top-performing prediction model in the competition. He is building on that success and leading the effort to create the real-time predictions for the upcoming flu season in the U.S.

The new funding for Reich and his lab also support other collaborations with the CDC, in particular with their dengue fever branch in San Juan, Puerto Rico. They are planning to extend this new prediction framework to dengue fever in Puerto Rico.

Another project for Reich and colleagues is funded by a two-year, $500,000 grant from the Defense Advanced Research Projects Agency (DARPA) that will focus on “putting biology to work for us and improving our infectious disease predictions,” Reich says.

One of the main challenges, he says, is “to figure out how to incorporate underlying biological factors that drive big outbreaks,” such as the severity of a particular circulating strain of flu or what percentage of a population already has immunity to a disease due to vaccination or previous infections.

He says, “DARPA understands that the most dramatic improvements to predictions will likely come from using basic biology to inform underlying susceptibility in a population. Using dengue fever as an example, if you have a population of people who have been exposed to dengue already or have been given any of the new vaccines (only a few of which are actually on the market), they will have a different immunological profile for the disease than others.”

The biology involved is not entirely clear, Reich notes, but “our team’s and others’ previous work about this has shown that for dengue fever, an individual experiences complex immune reactions after infection that first protect you but then may put you at risk to a more severe infection. We want to explore these phenomena and understand how it can help us predict outbreaks earlier and with more certainty.”