Science’s COVID-19 reporting is supported by the Pulitzer Center.
Many websites track the devastating spread of disease and death caused by the now-pandemic coronavirus, from the World Health Organization’s (WHO’s) global map to The New York Times’s tally of U.S. cases at the county level. But one of the earliest, an online dashboard run by Johns Hopkins University, has become the go-to place for the latest data on coronavirus disease 2019 (COVID-19).
With its black world map strewn with red circles and global, country, and state counts of cases, deaths, and recoveries, the Coronavirus COVID-19 Global Cases tracker sticks to the basics—no fancy graphs. Yet the site, which gets more than 1 billion hits a day, has become the most authoritative source for COVID-19 case data. It is used by news organizations and government agencies around the world. Its dashboard has been copied by states and countries. It has been spotted on a wall in a photo of the U.S. Department of Health and Human Services’s coronavirus war room.
Behind the site is Lauren Gardner, co-director of Hopkins’s Center for Systems Science and Engineering, whose previous work involved spatial modeling of epidemics of measles and the Zika virus. Gardner spoke with ScienceInsider on Friday, the day COVID-19 cases surpassed 1 million worldwide, with more than 50,000 deaths. This interview has been edited for clarity and brevity.
Q: There are a lot of sites tracking COVID-19 cases. How did yours come out on top?
A: Probably because it’s been around the longest. We started this in January when the outbreak was pretty much just in China. My grad student Ensheng Dong, who is Chinese, was personally interested. In a few hours, we built the original dashboard. And the next day [22 January] I shared it on Twitter, and it immediately became popular.
Q: The dashboard draws on hundreds of sources, from WHO data to sites that aggregate news stories and social media reports on COVID-19. How do you make sure it’s accurate and not double counting?
A: There are millions of eyeballs on it all the time. So, if we’re off, people reach out and contact us very quickly. We get thousands of emails. We’ll get told, “Hey, there’s two new cases here that you don’t know about.” We also now have an anomaly detection system in place that alerts us to discrepancies in the case reports that we automatically collect.
We do have to worry about loops [where our own data are fed back to us as original cases]. There is a media aggregation site for the United States called 1point3Acres that we follow really closely. We take U.S. data from them, and they pull global data from us. We have to be really careful to only reference their national data. But the thing is, if there is a loop no one’s reports [of COVID-19 cases and deaths] would ever increase. So, we know that’s not happening.
What I would like is for all the different local health authorities to keep improving their own reporting in a way that we can draw the data directly from them rather than from local media reports.
Q: Couldn’t you get the U.S. data from the Centers for Disease Control and Prevention?
A: You would think so. But they only provide state-level data, and it’s sometimes a 24- to 48-hour delay. There’s nothing at a county level.
Q: How big is your team?
A: At first, it was my group, which is about six people. But early on, Hopkins reached out and offered support internally. Because we were blowing up Amazon [cloud computing] servers with all the demand. Now, the Applied Physics Lab [at Hopkins] helps with the back-end data curation and tech. Esri, the company that has the mapping software, helps manage the platform. People at Hopkins manage the media and communications. But the group is still way smaller than it should be for what we’re doing.
Q: What is the workload like? Do you work in shifts?
A: For over 2 months, we were trying to make decisions on where to collect data from, what data was trusted, how to aggregate it, validate it. We initially did everything manually. Now, almost everything’s automated with various cross checks in place. The dashboard is automatically updated hourly. We’re also on a 24-hour rotating shift for things like server issues and data curation. For instance, we have a Ph.D. student based in England who gets our early morning shift.
It’s a big volunteer-based public service effort. We just are trying our best to make it be as good as possible, but we know it’s not perfect.
A: Yeah, every day is a new surprise. The geopolitical implications have been stressful and distracting. I just want to report the data that will be the most useful and appropriate for the people that are trying to get access to it. The virus doesn’t care about the borders.
Up until yesterday, we had a lot of cases without an assigned location on Null Island, right off the coast of Africa, which is [a location in the ocean that has] zero coordinates [latitude and longitude]. It’s famous. I thought it was a great place to put everything that doesn’t have a specific location yet. But that upset a lot of people, so that’s gone.
Q: As the number of COVID-19 cases grows, is it more work?
A: It’s actually less manual work now because it’s automated. We’re spending more time doing other types of research now. Almost 90% of my interests and efforts are back to mathematical modeling around this disease. We’re doing real-time risk assessment of what’s going on in the United States, aiming to feed these results back to policymakers to say, “Here’s the counties that we should be worried about tomorrow.” We can do it, so we should be helping with that.
Q: Are you getting much sleep?
A: It’s exhausting. We’ve been doing this full on since January. We dropped everything else in the lab. And it’s probably going to be this way for at least another couple of months. And I think we’ll be tracking the outbreak for a year. It’ll keep going and bouncing around all over the world. So, it’s a 110% effort for sure. I think all public health people working in this space feel the same.