Post-COVID, Systems Need to be Crisis-Ready for Better Public Health Response

When the COVID-19 pandemic hit the United States in early 2020, epidemiologists and structural biologists worked overtime to understand its pathology and speedily identify vaccines. Public health officials raced to develop metrics and strategies to address caseloads, outcomes and logistics.  Computational and data scientists developed novel systems capable of consolidating a sudden flood of information so decision makers could effectively govern as well as keep the public informed during a true public health crisis.

It was a complicated grind of urgent necessity, said Jonathan Ozik, principal computational scientist at U.S. Department of Energy (DOE)'s Argonne National Laboratory. It’s also one he doesn’t want to repeat.

“Let’s never do that again,” said Ozik.

The National Science Foundation (NSF) agrees. In the summer of 2022, it awarded Ozik, Argonne colleague Abby Stevens, and others, including researchers at The University of Chicago, Northwestern University, RAND Corporation and Virginia Tech, a $1 million planning grant for Predictive Intelligence for Pandemic Prevention (PIPP). The grant is given to high-risk, high-payoff research focused on addressing the prediction and prevention of infectious disease pandemics.

Previously, Ozik was part of a group that developed CityCOVID, a large-scale, data-driven infectious disease agent-based model that helped with epidemiologic forecasts and scenarios to support decision-makers.

With the PIPP funding, Ozik, Stevens and subject matter experts organized three workshops. Decision-makers and public health stakeholders who participated in the workshops revisited problems they faced during the pandemic and shared solutions they each developed. Common threads quickly emerged.

“Our workshops highlighted a recurring pattern of different public health departments and universities building unique systems from the ground up,” said Stevens. “Without a central resource, researchers were scrambling to set up their own modeling and analytics infrastructure to answer essentially the same questions as other groups and municipalities.”

For example, officials needed the ability to integrate the ebbs and flows of hospital visits, vaccine supplies, infection rates and mortality rates with epidemiological forecasting and scenario analysis tools. Without existing standard approaches, they had to create their own systems, more or less from scratch.

High performance computing resources, advanced epidemiological models and powerful algorithms could have helped. Ozik and Stevens believe such tools are essential to creating a suite of new resources at appropriate scale that can be applied across counties, states or the country in response to future crises.

“There are models, such as CityCOVID, that Argonne built and validated using advanced algorithms to understand how various diseases spread through populations,” said Ozik. “These models pointed to interventions that can be used to mitigate, and possibly prevent, the effects of novel infectious threats.”

In phase 2 of PIPP, the NSF is interested in funding national centers focused on understanding how to better support pandemic response and prevention. There, researchers like Stevens and Ozik could focus on developing proven, scalable tools that can be applied quickly in any future epidemic. For example, they could show how integrated public health responses, through robust collaborations between public officials and epidemiological researchers, can avoid the hectic scramble of the early days of COVID-19. Insights from the workshops are being used to inform such a potential future center.

“We’re looking to build scaffolding onto which researchers, analysts and decision makers could apply their subject matter expertise,” Ozik said. “This approach would be better than having everyone try to repeatedly figure out the various and complex details that require expertise across multiple, disparate domains to improve public health response.”

The first workshop brought together public health stakeholders and decision-makers from Illinois, California, and the Centers for Disease Control and Prevention. They sought to establish a shared vision for transforming public health response to emerging pathogens. Participants discussed co-design of policy, implementation and risk analyses, as well as robust data for modeling and prediction of future pathogens.

The second workshop adopted a One Health perspective. It brought subject matter experts in eco-epidemiological surveillance and public heath stakeholders together to develop a shared vision for the data and sensing capabilities needed for computational epidemiology. They looked at ways they might improve bio-preparedness by using rapidly multiplying data streams more effectively.

The third workshop examined the use of high performance computing and large-scale data management workflows to improve pandemic predictive intelligence. This workshop drew upon the experiences of the previous sessions and added the perspectives of computational and data scientists from national laboratories, academia and elsewhere.

“We want to understand what might be coming down the pipeline and create playbooks to respond well to fast-moving, novel threats,” said Ozik. “Researchers can’t be asked to drop their primary research, which is also essential to the public, and assemble in support of local decision-makers without the resources they all need to be effective. That's just not a sustainable way forward.”

Source: Argonne National Laboratory