Using supercomputers to tackle COVID-19

TACC's supercomputer saves lives with COVID-19 data modeling and pioneers a new era of pandemic science.

By Sara Downey, Thought Leadership, Dell Technologies

The Texas Advanced Computing Center, located at The University of Texas at Austin, just celebrated 20 years of powering discoveries with advanced computing. In that time, it’s undertaken a wide range of life-changing work, from creating models to predict how chemical weapons disperse in the atmosphere to identifying a new metric to predict blood clots in the heart and 3D imaging research to help develop climate-change adapted plants for farming.

To continue answering some of humanity’s toughest questions with petascale computing, the Dell Technologies’ powered Frontera (the fastest supercomputer on any university campus and the 13th fastest in the world), I sat down with Professor Kelly Gaither, director of health analytics at TACC and Professor Lauren Ancel Meyers, director of the University of Texas at Austin (UT) COVID-19 Modeling Consortium, both of whom use Frontera for their research.

Our conversation focused on TACC’s work, supporting researchers during the COVID-19 pandemic, in light of the strides researchers have made in understanding and mitigating this global emergency. Its first feat was the swift founding of the COVID-19 Modeling Consortium. Under Professors Gaither and Meyers’s humble leadership, the consortium was instituted within a month, with Meyers noting “our research went from zero to 60 overnight.”

To some extent, they were supported by previous epidemiological work undertaken by TACC. For instance, in 2009, TACC helped Meyers develop computational tools for preparedness during the H1N1 pandemic.

That’s not to say everyone immediately envisaged TACC and its supercomputer being a leading resource in the race against the virus. Gaither is aware that to the average layperson, supercomputers can seem like the stuff of science fiction, far removed from everyday problems. But TACC’s mainstream work with COVID-19 is changing that perception.

“We’ve seen sci-fi movies where you have this high-performance computing technology, and it’s used for lots of different purposes. And I think it makes people nervous sometimes. But here’s the other side of it—in times of crisis, we can preserve human health and safety,” Gaither said.

Saving lives with data modeling

At the outset of the pandemic, the greatest challenge the team faced was having to deal with the unknown, and specifically the dearth of data. However, the consortium responded swiftly by scraping data from 18 different public health departments across China. From that exercise alone, the discovery was “jaw-dropping”, according to Meyers.

“We used their case report data to infer how quickly the virus was spreading, and we discovered that people were spreading the virus even before they had symptoms. At the time, nobody quite believed that. Of course, this meant that the virus would be hard to control.”

Since then, they’ve accumulated copious data to drive forecasting and help people anticipate what might happen next. “Our forecasts help hospitals make sure they have enough resources to handle major COVID-19 surges. And they help decision-makers—from local community leaders up to the White House and even international organizations—make decisions that balance very difficult trade-offs,” Meyers shared.

The consortium runs dozens of models on any given day. From analyzing how to open schools safely to predicting how many people might be admitted to hospitals next week, the consortium’s data scientists are using the processing power of Frontera and other computing resources to equip cities and policymakers to decide when to trigger timely interventions.

In Austin, researchers analyzed data and built dashboards at the zip code level and below to power policy intervention. Meyers and her colleagues teamed up with Dell Medical School to assess the potential impact of the pandemic in the Austin-Round Rock area.

By understanding cell phone data and applying past epidemic growth trends, epidemiologists could not only track what had already happened, they could also create predictions for the future. This research has had life-changing results.

“When you look at the numbers in Austin, and you compare them to the numbers around the rest of Texas, you can see that decision-making based on good data and sound computational models has saved lives,” says Meyers. “Our death rate is less than half the statewide average, and much, much lower than the other major metropolitan areas.”

Moving forward with surveillance and swift responses

While the research continues to shape how cities and governments respond to the current pandemic, it will also mold how societies respond to and recover from future crises. In the coming years, Frontera will support rapid responses to emergencies such as earthquakes, tornadoes and other disasters.

Gaither noted that high-performance computing has evolved over time, and been designed with greater flexibility, which has enabled organizations like TACC to bring costs down as they scale up. “We’ve been able to continue to address extremely data-intense and critical problems, without losing financial feasibility.”

When I asked about lessons learned, Meyers responded: “We need to build more rapid and robust surveillance systems that track health and behavior around the globe, without infringing on individual privacy. Data-led approaches powered by supercomputing can help to slow transmission, save lives and contain future threats at their source.

“That involves not only bench work to identify viruses lurking in wildlife populations, but also collecting data on how people move and behave, so we can anticipate and prevent the future spread of a disease. It involves partnering with industry to build cutting-edge computational tools so we can rapidly analyze large volumes of data from around the world. When fighting a pandemic, we need to track risks and take actions on multiple scales all at once, from a single neighborhood to the entire globe.”

Gaither concluded, “Our work is pioneering a new era of pandemic science and will ensure we’re better-equipped the next time we see an emerging viral threat. The team and I feel incredibly privileged to be involved in such important work.”