Harvard Researchers Create Simulator to Help Policymakers, Public Understand COVID-19

Researchers at Harvard, Massachusetts General Hospital, and Georgia Tech have created an interactive online simulator to help policymakers better respond to the novel coronavirus.

After more than a month of development, the eight-member team released the simulator Friday. It features state and national projections of how different intervention scenarios would impact case counts, deaths, and healthcare capacity.

Project lead Jagpreet Chhatwal, a radiology professor at Harvard Medical School, said the biggest challenge the researchers faced was accounting for unknown variables — considerations that are becoming more apparent as the pandemic continues.

“We don’t know many factors, many variables, that will go in the model. So how do we really account for uncertainty? Many times, we had to change our approach,” he said. “Even going back from scratch and starting again as things are evolving.”

To create the simulator, the team relied on an SEIR model — an approach that depends on a system of differential equations to model shifts in susceptible, exposed, infectious, and recovered populations. Özden O. Dalgıç, a mathematical developer on the team, said they incorporated additional statistical tools to project the future course of the pandemic.

The researchers also acknowledged that the model does not take into account other potential interventions, such as vaccines and increased testing capacity.

Though Chhatwal is unsure if the simulator has informed state policy decisions, he said his team presented preliminary data to Monica Bharel, commissioner of the Massachusetts Department of Public Health.

HMS postdoctoral research fellow Peter C. Mueller, who previously worked with Chhatwal on a hepatitis C simulator, said small shifts in the timing of when policymakers impose or lift restrictions can have disproportionate effects on case counts.

Georgia, for example, reopened some businesses on Monday. Lifting restrictions over the next month, even gradually, could result in more than 23,000 deaths in the state, according to the simulator. Chhatwal and Dalgıç said their current model predicts that restrictions could be safely lifted by the end of June.

“This pandemic is just so contagious that every little subtle difference can make a huge impact,” Mueller said.

The team hopes to impart some of the most important takeaways from its research to the general public using the simulator, which has already garnered page views from all 50 states, according to Mueller.

“The two things that people should really learn from this is we weren't prepared enough to begin with, and we didn't act fast enough,” Mueller said.

Mueller said the team is considering a feature that shows what case counts would have been if restrictions were imposed sooner. Additionally, Dalgıç said, they plan to add the White House’s proposed phased reopening plan to the list of scenarios.

“Going forward, we're definitely gonna keep updating the data,” Mueller said. “Keep adding more projections.”

—Staff writer Simon J. Levien can be reached at simon.levien@thecrimson.com. Follow him on Twitter @simonjlevien.

—Staff writer Austin W. Li can be reached at austin.li@thecrimson.com. Follow him on Twitter @austinwli.