Skip to Main Content


Cross-Country Collaborators Monitor the Spread of COVID-19

Yale Public Health Magazine, Yale Public Health: Fall 2022
by Matt Kristoffersen


On January 20, 2022, during the height of the omicron variant’s winter sweep across the nation, a map of U.S. COVID-19 cases looks like a blizzard. 

Then, by spring 2022, as the county-level case counts fell nationwide, one can see a gradual thaw: A nation awash in light blue—what the map’s legend signals as startlingly high infection numbers—slowly switches over to the deep purple hues of a waning pandemic. 

It’s a colorful way of displaying epidemiological findings on a granular level. But to the creators of covidestim (a combination of “COVID-19” and “estimate”), including Yale School of Public Health Professor Ted Cohen, MD, DPH, it’s something far more valuable. The intense collaboration behind covidestim has produced valuable snapshots of the COVID-19 epidemic in real time, taking into account delays and incompleteness of surveillance data. 

We’re trying to estimate something that is not as easily seen. It’s an invisible quantity ... but it’s one of great epidemiologic importance.

Ted Cohen

Covidestim is a statistical model that the researchers developed with financial support from the U.S. Council of State and Territorial Epidemiologists. It runs with support from Amazon and the Yale Center for Research Computing. It does what the researchers call “nowcasting” of the pandemic. To see the latest covidestim model, visit

“Our goals with this project are focused on developing as granular as possible a picture of the current state of the epidemic, given all the limitations with the available data,” said Cohen, a professor of epidemiology (microbial diseases). 

Unlike popular COVID-19 case counters like the Johns Hopkins COVID-19 Dashboard, covidestim estimates the numbers of infectious individuals regardless of whether they are symptomatic or have been tested for disease. It works by combining county- and state-level observations of case numbers, hospitalizations, deaths, vaccination rates, and existing data on the natural history of the virus to estimate the numbers of unobserved infections that have occurred in a specific location. The estimates that the team generates are accompanied by uncertainty bands. 

“It’s important to quantify and communicate our current levels of uncertainty because infections happening today result in clinical cases, hospitalizations, and deaths in future weeks. As a result, estimates for what’s happening right now or in the very recent past are imprecise,” Cohen said. 

Cohen credits his collaborators, especially the two scientists with whom he has worked for more than a decade on other projects—Harvard Associate Professor Nick Menzies and Stanford Professor Joshua Salomon— with helping to make covidestim a reality. 

The cross-country teamwork has definitely been beneficial, Menzies said. 

“Having a foothold in three different institutions in three different states gives us access to a wider set of experts and decision-makers to work with,” he said. 

The open-source nature of the project is one of its greatest assets. Anyone can freely download covidestim’s code to generate models with their own data. This allows researchers to modify and track the path of other pathogens as needed. 

“We didn’t want to set ourselves up as gatekeepers,” Menzies said. 

Since 2020, the covidestim project has helped researchers, decision-makers and members of the public health community understand where COVID-19 was spreading, Cohen said. The “nowcast” team produces helps with that process. 

“We’re trying to estimate something that is not as easily seen. It’s an invisible quantity,” Cohen said. “But it’s one of great epidemiologic importance.”

Next Article
YSPH Partners with Connecticut Department of Public Health on Workforce Training Programs