One size does not fit all when it comes to pandemic precautions. So say three local researchers who are developing and testing a computer model that will better guide health officials in the future.
“Most people agree that we should have and we could have done much better,” says Subramanian Ramakrishnan, a University of Dayton associate professor of mechanical and aerospace engineering. “There is a critical gap in terms of the fundamental knowledge about how an epidemic spreads.”
Ramakrishnan, University of Cincinnati mechanical engineering professor Manish Kumar and Cincinnati Children’s Hospital medical doctor and epidemiologist Shelley Ehrlich have a National Science Foundation grant to come up with a suite of models that will learn as they go.
“We come from two different worlds really, but sometimes it takes that to kind of come up with some interesting ideas to solve problems like the COVID pandemic," Ehrlich says.
What makes it so difficult to determine spread?
Two factors contribute to the uncertainty of how viruses spread:
- Random human behavior (deciding whether to wear a mask or social distance)
- Random motion (traveling in neighborhoods or cities)
Their model learns from health department data and tries to predict the spread 15 days out, based on human behavior. Then, it compares the prediction to what actually happened. Ramakrishnan says you can’t really take a mathematical model and expect it to satisfy the conditions for a much larger area. That’s why the system works best regionally.
“You probably want to focus locally or even regionally, say the Tri-State, for example,” says Ramakrishnan. You can “use these models to get a sense of how it’s spread and how it can be controlled.”
This makes sense because people act in different ways in different parts of the country.
Early in the pandemic in April 2020, the team took data from the state and the Hamilton County Health Department to see how well their initial model would perform. They say it did well for Ohio but even better in Hamilton County.
“One part of the project will be to really validate this model against existing epidemic models to see how it performs better, which we expect it will,” says Ehrlich.
