In January, as COVID-19 began its spread around the globe, a team of Johns Hopkins University engineers began tracking data from around the world to document the novel coronavirus. Their online dashboard now gets upwards of 4.5 billion hits a day from public health authorities, researchers, and the general public. In early May, we spoke with project lead Lauren Gardner, an infectious disease specialist and co-director of the school’s Center for Systems Science and Engineering, about the dashboard’s viral moment.
What was a normal day like for you before the coronavirus outbreak?
A lot more relaxed. My group does infectious disease modeling, specifically looking at the risk of spread on national or global scales. We look at how many different factors—people, their environment, land use, population density, urbanization, climate mobility—play a role to understand which are most significant, then we build out prediction models. Back in January, we were looking specifically at measles. I’ve historically done a lot of work around Zika and dengue fever.
Given your expertise, did the arrival of this pandemic come as a surprise?
Back in January, I definitely didn’t think this was going to become the worst pandemic in a century, but in general, I am keenly aware of the heightened risk of new novel viruses getting introduced into the human population. The severity of this one, the extent to which it has integrated into every part of the world, and the complete economic collapse around it, is more extreme than I, and I think most people, were originally thinking. Even in January, we were expecting to see it all over the world—we were modeling importation and exportation risk at a global scale back then. But there was still just so much uncertainty surrounding how infectious it really was.
How did you make the shift to following COVID?
We had a lab meeting back in January with a couple of my PhD students and we were talking about this new coronavirus in China, just discussing what was going on and the implications. The students were already tracking it closely because they’re both Chinese and have family there. I thought it could be a great opportunity to build out a dataset for an emerging infectious disease in real time for the research community. Of course, we didn’t know where it was going to go. We built the dashboard just to visualize the data. It was a big surprise how popular it became for the general public. It makes sense now. My PhD student, Ensheng Dong, was really the pioneer. He built the original dashboard from scratch. We literally made the whole thing in a few hours one evening. But the data collection feeding into it is an incredibly complex process that has massively evolved over the last few months.
How quickly did it begin to consume your work?
The first week we put it up. It was growing quickly in popularity. By the end of February, it was already getting a billion hits a day. It was huge, and already starting to quickly get unsustainable in terms of managing it. We had to grow the team and start automating things. Part of it too was the scale of the outbreak. In terms of magnitude and also spatial scale, it just grew so fast through January. It was pretty obvious very fast that we were getting in deep.
How timely and accurate is the dashboard?
We update every hour. It just depends on how frequently our sources update, and because we’re reporting from all over the world, we’re getting updates at different times. One of the biggest challenges has been that sources are changing all the time. Over the last few months, counties, states, and even countries have come up with their own reporting systems, dashboards, and websites dedicated to COVID. None of these existed in January and February when we first started tracking this. Reporting rules are also changing. Some places include probable cases, some include probable deaths, some don’t. That’s really complicated to keep up with, and it’s a reason why there are discrepancies. Everyone in the world is still trying to learn how to deal with this, and their data collection and processes are continually evolving. Even in the U.S., there are inconsistencies all over the place.
What’s your day to day like now?
Since January, it’s been 110 percent. None of us have taken a day off. We work pretty much all the time because the dashboard is running all the time. We have people on 24-hour rotations for issues around servers and scripts and data crashes and all that, and we have group meetings every morning to talk about any ongoing issues, because we’re still literally building out the infrastructure every day. On the side, I’m doing a lot more modeling and analysis now, too, using the data we’ve collected through the dashboard. We’re creating weekly risk assessment forecasts for what’s going on in the U.S. We’re looking at how the cases and death rates are growing in every U.S. county. We’re looking at how social distancing is affecting the outbreaks growth rates. We’re looking at building local and global models for climate seasonality, and how that’s affecting COVID around the world.
Are there any key takeaways?
We’re learning a lot. The most critical thing right now is that we’re obviously seeing the loosening of regulations around social distancing and stay-at-home orders, and I would say it’s definitely not time for that yet. There’s strong evidence that social distancing is driving down the growth rate of the outbreaks, and if we start moving around again, we should expect to see those increase.
Are there any local trends you’re paying attention to?
A lot of our county-level modeling looks at the contributing risk factors, and a lot of that has to do with certain socioeconomic demographic characteristics, so I definitely have concerns about Baltimore. I do expect it to get worse here. Maryland is one of the higher risk states that we’re concerned about.
What’s next for the dashboard?
We’ll keep it running and continue to build on it and keep it as sustainable as possible. We’ll continue to create new models to try to learn more about COVID, and then ideally we’ll be able to transfer that knowledge to other infectious diseases. I think the dashboard has filled a knowledge gap and set a great example in terms of having a public facing tool to watch these things unfold in real time. I think it could be a good resource for annual influenza. And I hope it encourages more vaccinations among populations because maybe they’ll understand more about the risks.
You joined the Hopkins team in early 2019. What it’s like being a part of that community in this moment?
This tool is so trusted because it’s coming out of Hopkins. I think it’s really cool that we’re doing it out of the [Whiting] School of Engineering. It’s important for people to recognize that engineers can contribute to society in lots of different ways, including public health.