Epidemiological modeling

Model behavior

ASU computational modelers unleash the power of mathematics and data science to help Arizona conquer COVID

From your smartphone’s weather app telling you it will rain on Friday to the check engine light in your car to the safety of the airplane you’re about to board, computational modeling touches your life every day.

But what exactly is it? Combining the power of mathematics and data science, computational modeling replicates and forecasts real-world events using computer simulations. The computers return possibilities about what is happening — or could happen — in complex situations, from weather forecasting and aircraft flight to earthquakes and pandemics.

At Arizona State University, computational modelers work with researchers in a variety of disciplines to create useful models and interpret their results. They help organizations and individuals make informed decisions in uncertain situations.

Unleashing data to curb a pandemic

Few situations are as uncertain as a global pandemic caused by a completely unknown virus. As COVID-19 surged in Arizona and throughout the country, so did people’s questions. How does the virus spread? What is the best way to protect ourselves?

As the delta variant of the virus emerged, new questions have arisen. Are breakthrough cases now the norm? Do we need booster shots, and when? Without concrete answers, public policy makers, health care leaders and corporate executives have struggled to formulate plans about reopening strategies and safety policies.

Filling the information void is Arizona State University’s COVID-19 modeling task force, Modeling Emerging Threats for Arizona (METAz), led by Timothy Lant, director of program development at ASU’s Biodesign Institute. The task force includes Megan Jehn, an associate professor in the School of Human Evolution and Social Change who teaches courses in epidemiology, global health and quantitative research methods; Anna Muldoon, co-author of “COVID-19 Conspiracy Theories”, research professional at the Biodesign Institute, and graduate student in the School for the Future of Innovation in Society; and Heather Ross, an ASU clinical associate professor who holds a joint appointment in the School for the Future of Innovation in Society and the Edson College of Nursing and Health Innovation.

These experts in epidemiological modeling have emerged as leading providers of data about the coronavirus and its rapid spread in the state, helping government agencies such as the Arizona Department of Health Services, businesses and individuals make informed decisions on how best to protect human health and well-being.

Saving lives with modeling

“One of the big things we focused on with our models was hospital capacity,” Ross says. “We were able to let public officials and health care officials know what they would need in terms of COVID beds and nurses to care for patients. Hospitals were then able to make adjustments, such as cancelling elective surgeries, and the public stayed away from the emergency room for things that weren’t true emergencies".

The team also warned of the dangers of eliminating mask mandates too soon as the state granted local governments autonomy to establish their own health and safety measures to fight rising COVID-19 cases last summer.

We were able to make sure that everyone who needed an ICU bed in the state of Arizona was able to get an ICU bed.” — Clinical Associate Professor Heather Ross

“I’m confident that our model helped some public officials retain their masking policies,” Ross says. “And we know that those masking policies helped save some people’s lives.”

“Similarly, we were able to raise the alarm around Thanksgiving, when our epidemiological model showed that household gatherings where people were unmasked in closed spaces would lead to an even bigger surge,” she adds. “And, indeed, that’s exactly what happened.”

How does epidemiological modeling work?

Epidemiological models, like other computational models, “speak” in the language of mathematics. They are used to track infectious diseases in populations, identify the most effective interventions and monitor and adjust interventions to reduce the spread of disease. At ASU, METAz team members update data that represent a set of conditions, such as different vaccination rates, or rates of transmission depending on mask policies, social distancing and other behaviors. The computer runs these numbers through a set of equations to simulate what would happen in the real world under these conditions.

The computer spits out even more numbers in return. The scientists use software to take those numbers and turn them into graphs, maps or other images that help them understand the results. The final step is to explain those visualizations in the context of current conditions, ranging from policy environments to changes in the virus’s biology, like the emergence of variants.

“It’s a process of looking at data that’s available at any point in time and trying to infer what’s going to happen next, which is a prediction question, and whether there are interventions we can apply that would change the trajectory of the disease, which is an intervention question,” Lant says.

Before the pandemic struck Arizona, METAz team members had emerged as experts in diverse fields such as mathematical epidemiology, projection modeling, management and decision science, public health systems, machine learning, health policy, medical and nursing science, and more. Their deep subject matter expertise made it possible for them to spring to action to create reliable and trustworthy models at a time when the state’s public health policymakers and health care decision-makers needed them most.

A glimpse into the future

ASU’s pandemic modeling efforts revealed important lessons for the future, such as the need for a better public health data infrastructure for public health officials to identify hotspots and inform reopening decisions.

“We really need to close some of the gaps to get better at our public health surveillance and reporting at the state and county level so that we have real-time data that’s feeding these models to make them more accurate,” Jehn says.

Meanwhile, pandemic modeling tools and data dashboards developed at ASU will prove to be as useful in the future as today. Lant reminds us that “we really have no reason to anticipate that COVID-19 is going to go away, because there’s so much evidence to the contrary.”

What’s next? METAz is now taking a look at the future “endemic” phase of the coronavirus, when the virus will be continually present but affecting a relatively small number of people.

As scientists hone more powerful algorithms for disease forecasting, Lant foresees the day you’ll be able to check an app to monitor infectious diseases in your community — much like using a weather app that predicts sunny skies, rainfall and humidity.

“The National Weather Service has evolved into a global system of weather forecasting that includes sensors, computer models and high-performance computing infrastructure,” Lant says. “Researchers have taken science and brought it to the point that we get weather reports in our homes by watching TV or checking an app. Likewise, the understanding of infectious diseases could advance to the point that we could check an app to see how much disease is around.”

See the original article on ASU News Oct. 22.

Written by Lori Baker, communication specialist, Knowledge Enterprise. Image by Jason Drees