It’s been nearly three years since the pandemic that took millions of lives officially ended, but Northeastern University network scientist Alessandro Vespignani and his team are not letting their forecasting tools gather dust.
When COVID-19 caught the world off guard in early 2020, the Northeastern scientists developed transmission models that predicted pandemic levels of disease by March of that year.
Now they have created post-pandemic datasets and maps of what Vespignani called the first study of how Americans interact and move in the post-COVID era. It comes with a mobility platform that serves as a real-time dashboard on population movement, as well as software — called Epydemix — that allows scientists and public health experts to plug in their own data to compare disease transmission modeling scenarios.
“During COVID, we built tools while using them,” said Vespignani, Sternberg Family Distinguished University Professor and director of Northeastern’s Network Science Institute, who compared science efforts to model disease transmission during the pandemic to building a plane while flying it.
“Now COVID is over and we are going back to normal,” he said. “Many of those technologies, approaches are being abandoned. We cannot do that. We need to keep those tools sharp and ready for the next time,” even flu season.
Vespignani said the technology his team unveiled earlier this month helps create the type of situational awareness that guides public health officials in making informed decisions about interventions, whether they be vaccination campaigns, keeping schools closed — or open — and hospital staffing.
“We don’t sit on the data. We share with our partners and make the data publicly available on our web pages,” he said. The open-science project is part of the mission of EPISTORM, a CDC Center for Forecasting and Outbreak Analytics (CFA) funded initiative led by Vespignani since 2024 that is focused on improving early detection and preparedness for infectious disease outbreaks in the U.S.
“The idea is that you can actually use this information to modulate transmission,” Matteo Chinazzi, research associate professor at Northeastern’s Roux Institute, said about the mobility platform dashboard.
The software Epydemix is intended “to lower the barriers between the people who actually need the tools and the practical implementation of epidemic models,” said Nicolo Gozzi, research scientist collaborating with the Northeastern Network Science Institute on the project.
The importance of population movement
The post-pandemic behavior data uses GPS movement data from more than a million mobile devices, anonymized to protect privacy, to discern how people interact and travel to different locations and point of interests to better understand infection risk, Vespignani said.
He called contacts the “wiring” that turns infections into outbreaks — and said that wiring did not snap back to pre-COVID times when the emergency ended. For one thing, people now have fewer contacts in the workforce.
The ongoing data updates from EPISTORM can be plugged into different infection scenarios to determine, for instance, whether to close schools and whether mandated policies are required or if the public is already changing behavior due to disease awareness, Vespignani said. “If we keep using old contact assumptions, we will misread transmission risk and mis-time preparedness,” he said.
A real-time dashboard on mobility
Knowing how populations move is key information for large-scale epidemic models to predict where and when disease will spread.
The U.S.mobility platform developed under EPISTORM offers real-time measurements, updated monthly for now and weekly, eventually, Chinazzi said.
“When COVID started, we didn’t have access to this kind of data, so we had to build it from scratch,” he said. Other companies, such as Apple and Google, started releasing similar products but discontinued them once the pandemic was over.
“What we’re trying to do is fill that gap and have (the mobility data) always live and ready,” ready for the next flu season or emergent outbreak, Chinazzi said.
The mobility data measures how far devices move from a center of mass, like someone’s home, across a typical day and during holidays or special occasions like the upcoming Mardi Gras celebration in New Orleans, he said.
It also measures contacts such as when two devices are close in a physical space. “We can actually measure not only the number of distinct contacts per user, but also the duration of the contacts and the average duration,” Chinazzi said.
Accessible software
The accessible software provided by Epydemix allows sophisticated epidemic modeling historically restricted to specialized research teams to be used by public health officials and smaller research departments, Vespignani said.
It’s called a “no code” open-source toolkit, Gozzi said. Epydemix was born as a Python interface, but even those who don’t know how to code in Python can use it through various built-in dashboards, he said.
“Imagine a website where you don’t have to write any code. You just define your model,” run various scenarios using real-world population data and epidemiological data, such as how contagious the disease is and to whom, Gozzi said.
Scenarios can include what happens if workplaces or schools remain open or closed, he said.
Shoba Nair, director of epidemiology and evaluation for the Boston Public Health Commission, whose staff attended an EPISTORM training on Epydemix, said the platform will be useful for forecasting how different factors, such as vaccination levels, could impact the trajectory of infectious disease outbreaks.
“While the current package is primarily a tool for state and national forecasting, we are looking forward to working with our EPISTORM partners” to adapt the package to the city and local level, she said.
COVID-19 created an emergency call for contact tracing and mobility data, Vespignani said. Now EPISTORM is creating infrastructure and handing out tools and data, and capacity to policymakers.