Georgia Forecasts Hospital System Capacity during COVID-19 Pandemic with Data Analytics Software

With more than 8 million confirmed cases around the world and more than 2 million in the United States, the coronavirus disease 2019 (COVID-19) is an unprecedented pandemic that has compelled federal and state governments to implement rapid emergency response. From dashboards reflecting local statistics to daily press briefings and drive-through testing sites, each state has had to manage its response to the pandemic and create innovative solutions to protect residents.

The State of Georgia is no exception, coping with increased hospitalizations and unexpected demands on the health-care system due to 60,000 plus confirmed COVID-19 cases in the state. To evaluate hospital resources and anticipate what might be needed moving forward, state leadership in Georgia, including the Governor's Office and the Department of Public Health, needed a way to better understand, forecast, and visualize critical hospital resources necessary to support COVID-19 patients.

A new task force under the direction of Brigadier General Emmanuel Haldopoulos, chief of staff with the Georgia Air National Guard, utilized analysis software to develop forecasting models and determine whether or not state officials should increase the health-care capacity of state hospitals to aid hospitalized patients with COVID-19. 

To better evaluate hospital resources in the state, Haldopoulos assembled the Georgia COVID data analytics team. Georgia governor Brian Kemp requested data to determine whether or not officials should surge health-care capacity in Georgia based on demand forecasting (using past data to make future predictions) in response to COVID-19 hospitalizations. With the rapidly changing crisis, the governor wanted the data in less than 26 hours.

"The only question I was tasked with was, Should we or should we not spend millions of dollars to increase the capacity and the capabilities we have in the Georgia health-care system and bring in outside support?" Haldopoulos says.

Susan Miller, State of Georgia geospatial information officer and a task force member, says state leaders needed to better understand existing capacity and surge capacity, defined as the ability of a state hospital to rapidly and independently raise the number of resources available to support emergency events.

"We needed a better understanding of what beds we have—general and intensive care unit (ICU) beds—and ventilator capacity. We needed to understand the baseline for all of these resources and then immediately start exploring what the potential onslaught of patients might look like around the state under different scenarios," says Miller. "State leaders needed to understand current capacity, as well as predict what the near-term future might hold in store for the state during this unprecedented crisis."

She adds, "They wanted to be as prepared as possible to fill any potential gaps that might occur due to the outbreak to help save lives."

The task force needed a new solution that would allow state officials to easily view data and make fast decisions. He explains that to assess capacity, a database other than a mix of spreadsheets and a process for forecasting had to be developed to allow hospital administrators to report current status.

"We were able to advise [Governor Kemp] with very rudimentary data and graphics that were spreadsheet based, but we knew that the need—not just for that decision but others that would be asked of us in the future—would require some real professional visualization. And that's when we expanded with Esri," says Haldopoulos.

To begin this predictive modeling on hospital beds and ventilators, Miller says they assembled a collection of amazing minds from different institutions, including Georgia Tech and the Massachusetts Institute of Technology, and utilized data from the University of Pennsylvania's COVID-19 Hospital Impact Model for Epidemics (CHIME) as well as the University of Washington's Institute for Health Metrics and Evaluation (IHME) model.

Miller says, "What we ended up with was an absolute dream team of amazing scientists and subject matter experts from a number of fields all focused on helping us answer critical questions that ultimately enabled the state to be better prepared for the coming weeks and the spread of the disease."

The new task force selected ArcGIS Insights for its capability of translating results into visualizations. ArcGIS Insights is an analytics workbench that offers spatial and nonspatial analysis capabilities and allows users to perform iterative and exploratory data analysis.

Miller says Insights is just one of the Esri tools they implemented to visualize the reality of what's happening in state hospitals and predict the possible impact of COVID-19 under different scenarios.

"We chose Insights because we needed a single place to bring copious amounts of disparate data to life and empower leadership to make data-based decisions. We needed to be able to look at a variety of combinations of the models and break them down by geography to better understand current and surge capacity along with predictive numbers," says Miller.

She adds, "This would allow leaders to better anticipate where we might have surplus or shortages of ICU beds and ventilators needed to save the lives of our residents who become deathly ill."

To forecast what the possible peak demand on the Georgia health-care system would be, the task force focused on the aforementioned critical assets and broke these down beyond the state level to the regional coordinating hospital level, Haldopoulos explains. There are 14 across the state.

Esri published a forecast model template/script that was immediately usable in Insights, which the task force used to run its models along with Esri models and visualize them in Insights. The team used Insights to see all the different models together in one view to help them consider the different approaches and the many data points to help forecast needs for the state.

"With little guidance from the principals beyond the context of some dialogue about the spreadsheet chart and its intended objectives, the Esri team took a 'high school athlete and produced a Navy SEAL,'" Haldopoulos says. "I'll never forget when I first saw the initial draft presentation of the capacity planning tool. I was truly amazed at how the team was able to capture nearly every aspect of the intended functionality."

The use of Insights has given the task force the ability to quickly and efficiently provide a visual of the outputs of various models in one place so it's easily consumable by decision-makers. Haldopoulos says these visuals have helped state leaders make informed decisions faster during the COVID-19 pandemic.

"Anytime a policy authoring entity is confronted with decisions made complex by multiple variables, it helps tremendously to graphically depict a situation," Haldopoulos says. "Visualizing the efforts of a data analytics team for state leadership aids in the delivery of the information, both in time and the understanding of the material."

ArcGIS Insights has made data collection and analysis much faster for the team, and the team says the difference has been night and day. Haldopoulos says he was spending half a day gathering data from the five modeling groups and trying to display it on a spreadsheet in a "consumable format by which the pivot table could be populated and the charting created to depict in even the simplest manner."

Miller echoes this sentiment, saying, "Once we got our hospital capacity planner (our Insights tool) built, General Haldopoulos was able to visualize all five models simultaneously for all variables and resources. So, the functionality excelled in every way. The time savings is astronomical."

The ability to create visualizations of data for state officials has been incredibly valuable during the unfolding crisis. "I sat in a room of leaders, and I watched them use the visualizations in Insights to help draw conclusions and make decisions and was so thankful we could give that to them. The analytical and visualization capabilities are incredibly powerful," says Miller.

The task force has received positive feedback following the implementation of ArcGIS Insights. "The organizations we're working with have all repeatedly commented on the task force's excellent work visualizing the observations of these data analytics teams and believe it to be a collaboration and product unmatched in the nation," says Haldopoulos.

Miller adds, "The amazing Esri team made this possible through their technical and subject matter expertise, their industry knowledge with geospatial and epidemiological sciences, and most of all their dedication and passion for helping others in time of crisis."

In an unprecedented crisis, the insights revealed from this project have better-equipped state leaders to make critical decisions that can help save lives.

"The use of the Insights tool has literally brought data to light for multiple stakeholders who would've never had access to such insights and the results of our modeling teams' work," says Miller. "It enabled state leadership to better understand the current landscape for hospital capacity under COVID-19, thus enabling them to be better prepared for the outbreak in Georgia."