Using computational modelling to understand and respond to COVID-19 outbreak

They all have a name; the best supercomputers do. Currently, the world’s most powerful is in Japan – it’s called Fugaku. It has three times the power of the previous heavyweight, the Summit supercomputer at the US Department of Energy’s Oak Ridge National Laboratory, and more computing power than Sierra – or ATS-2 – a supercomputer built for the Lawrence Livermore National Laboratory for use by the US Nuclear Security Administration.

Simon Fraser University has a supercomputer, too. It’s called Cedar. It’s one of the most powerful academic supercomputers in Canada, and is paving the way for some fascinating new research breakthroughs.

One can only imagine what Chaitanya Kaligotla could do with it.

Because Kaligotla is a social scientist who specializes in computational modelling of human and social behaviour, at scale. He uses advanced computational and analytical methods to study how people behave in large complex systems (think healthcare interventions, misinformation spreads), and has traditionally applied his skills to help predict answers to big ‘what if?’ questions in the public health and information domains.

And nowadays, most of those questions centre around Covid-19.

Kaligotla already has experience in this domain and recently co-authored a paper called ‘Development of large-scale synthetic population to simulate Covid-Transmission and response’ (2021)

Supercomputers help answer ‘what if?’ health concerns

Partnering with the University of Chicago, Kaligotla was part of a team at Argonne National Laboratory that created a model called CityCOVID. It is a city-scale, agent-based model that mapped the daily actions and behaviours of millions of citizens in the greater Chicago region. Using Theta and Bebop, supercomputers hosted at Argonne National Laboratory, the team leveraged publicly available information from the national Census and census-based datasets as well as anonymized cell phone location data (from a company called safegraph) – an extremely robust and large data set that generated more than 8.54 million synthetic agents across 5.15M locations – to run tens of thousands of simulations on a synthetic population that statistically represented 7 counties in the Illinois region (that includes the City of Chicago)

Described by some as “SimCity on steroids”, CityCOVID simulations continue to be regularly shared with the Chicago Department of Public Health, City of Chicago, Cook County as well as Illinois State officials to aid in decision making.

Importantly, it afforded the team the opportunity to explore the potential impact of the virus and help make sense of some very complex questions to help state and city level decision makers, including:

• Will there be a second (or third, or fourth) wave? When and how bad will it be (in terms of expected hospitalizations and deaths)?

• Are some communities and neighborhoods being disproportionately affected?

• What is the effect of individual behaviors (mask wearing, social distancing) on the spread of COVID-19?

• Are some ages or race of people being disproportionately affected?

• What is the impact of school reopening strategies on the spread of COVID-19 throughout the community? Restaurants? Churches? Retail stores?

Building on what we’ve learned

Now that Chaitanya Kaligotla has joined faculty at the Beedie school of business, the same thinking could be applied to Metro Vancouver and the lower mainland region. The same model could be executed to better understand the granular and aggregate social behaviours of BC’s residents . More, since SFU’s Cedar supercomputer has the capacity to do so, a synthetic population for Canada could be created to map the transmission of the virus across the country. Is there a specific segment of our community more vulnerable than others? is it safe to travel between provinces? What are the impacts of the vaccine program? What are the potential impacts of a vaccine passport? Computational Modeling and Simulation methods could provide answers and direction across a myriad of issues.

In truth, there is no end to the number of applications for high performance computers. Kaligotla’s work to date has focused on using computational methods to map how the virus is transmitted, but the same methods and machines are also used to take a closer look at the virus itself, how it mutates and when best to prescribe drugs to promote resistance. SFU’s Cedar, and others like it, can also be used to better understand earthquakes and forecast hurricanes, as well as predict the impacts of climate change. With the power to simultaneously run hundreds of thousands of calculations, supercomputers will undoubtedly also play a major role in not only understanding pervasive social or medical issues, but also predicting outcomes to complex theories and solutions for many years to come.

Chaitanya "CK" Kaligotla is an Assistant Professor at the Beedie School of Business, Simon Fraser University. He teaches BUS 462 Business Analytics