Researchers at the Biocomplexity Institute received a $10 million “Expeditions in Computing” grant from the National Science Foundation on March 25 to use computational and engineering methods to answer fundamental questions about epidemics and pandemics in real time. Although the team has been focusing on the COVID-19 pandemic since late December, the five-year grant aims to answer general questions about how these crises arise and model the best steps to take to minimize transmission and allocate resources.
With the data being collected globally, the Biocomplexity Institute has begun modeling the success of various governments’ policies, projecting which health systems will need the most resources in the near future and where the next virus hotspots will be. The Institute also works closely with federal, state and local policy makers, helping decision making related to the pandemic by sharing the projections they have modeled so far.
The grant also supports the group’s extended network of 14 institutions nationally and over a dozen researchers internationally to collect data from the ground. Madhav Marathe — computer science professor and head of the Institute's research team — said that partners in different communities help researchers understand what interventions are taking place around the world and how people are behaving in the face of increasing restrictions.
“Epidemics do not obey national boundaries,” Marathe said. “The idea is that if an epidemic starts in a different country, then the local scientist on our team would take the lead in … identifying the questions that are relevant for that epidemic in that part of the world, and then guide the team to answer questions or build tools that would be relevant to the particular outbreak.”
Marathe thinks that we are still in the early stages of the COVID-19 pandemic. While the crisis loci might vary with time, numbers of transmissions will continue to rise. Interventions put in place to increase social distancing — including the decision to move University classes online — are useful for slowing down the epidemic, he explained, but the fight for preventing transmission must continue to prevent flare-ups in later stages.
One challenge facing this project is the lack of reliable, timely data. Researchers at the Institute and their national and international partners collect data from federal departments like the CDC, state health departments and even social media. Marathe explained that because pandemics do not happen very frequently, there is very sparse data available compared to what is needed for complex modeling. While gaining accessibility to more data would be helpful for the models, Marathe emphasized that it should not be at the expense of individual privacy.
Another challenge is knowing what to leave out of models.
Stephen Eubank — a professor of public health sciences in the Biocomplexity Institute — said that much of the groundwork for making predictions during a pandemic is understanding the mathematics of different models, including deciding what data is irrelevant. In the case of COVID-19, there is still little information about the disease, like what factors can increase the likelihood of infection and how many people are asymptomatic.
“The thing that's hard has been not understanding much about the disease itself, the illness itself,” Eubank said. “If we have a sense of where the virus is, at any point in time, then we can make projections ... and they're particularly good for being able to say what the effective interventions are.”
Eubank said that fortunately, the Biocomplexity Institute had previously been involved with computational epidemiology during other crises like the 2009 influenza epidemic. While COVID-19 is not the flu, it spreads in a similar way so researchers can use previous models to study virus transmissions. However, the long-term aim of the NSF grant is to create a system that gathers data from all over the country and the world and can then answer fundamental questions to help quell the pandemic, such as how many people are infected, what the rate of hospital admittances is and whether people will follow guidelines from authorities on how to mitigate transmission.
Anil Vullikanti, a professor in the Institute and computer science department, agrees that the overarching goal is to be able to quickly answer those vital questions when future pandemics occur.
“The broader goals of this grant are to develop tools for real-time epidemiology,” he said. “For the next pandemic, we hope to have tools for collecting data, for developing models, calibrating them, making predictions in a much more efficient way than we are doing right now.”
Vullikanti is currently investigating the burden on health systems and ways to minimize that load. By understanding how to make the most efficient use of the resources at hand, the projections from these models could help policy-makers prepare for and react to future outbreaks. The Biocomplexity Institute has created a dashboard with over 1 million visits daily that contains updated data related to COVID-19 from all over the world. While many of the data sets are publicly available, Vullikanti said that some sets come with restrictions and can therefore not be released to the public.
The questions that are at the core of the “Expeditions in Computing” grant are complex and interdisciplinary, because epidemics not only play out in bodily disease but political, social and economic changes.
“Epidemics are not just a health phenomena,” Marathe said. “That's how people studied it for a long time. But as you can see after COVID-19, it's a very complicated interplay of disease biology, which is how diseases spread, physiology and immunology, how bodies interact, but also social sciences, how people behave, and economic sciences.”
Christopher Barrett — the executive director of the Biocomplexity Institute — agrees. The Biocomplexity Institute as a whole is “interested in how information is born and transmitted in living systems,” Barret said, and that research involves multiple pieces from various disciplines.
“For an example … public policy associated with locking down and how that affects the economy is ultimately being driven by mutation mechanisms in the coronavirus with respect to our immune systems,” Barrett said. “We work on real problems and real problems don't know what department they live in.”
Although the Institute is not solely interested in researching pandemics, the spread of COVID-19 has led the Institute to focus on this topic. Barrett echoed that the world is only in the beginning stages of this pandemic, and its severity demonstrates the importance of studying these fundamental questions of computational epidemiology.
“As bad as this is, it's the early stages and we had better prepare ourselves,” Barrett said. “This research was proposed … long before this event occurred, which means there is an awareness that this is important stuff to worry about, think about and do serious work with.”