• New York City was one of the early hotspots for the spread of COVID-19. New York University researchers funded by a grant from the National Science Foundation have been studying human behavior near medical facilities to help inform policies on pandemics and other potential disasters.  GagliardiPhotography/Shutterstock
     New York City was one of the early hotspots for the spread of COVID-19. New York University researchers funded by a grant from the National Science Foundation have been studying human behavior near medical facilities to help inform policies on pandemics and other potential disasters. GagliardiPhotography/Shutterstock
  • Many people leaving medical centers during the COVID-19 pandemic went to crowded public spaces, such as the New York City subway system, potentially spreading the disease. Researchers hope studying such behaviors will allow officials to implement effective policies.  Clari Massimiliano/Shutterstock
     Many people leaving medical centers during the COVID-19 pandemic went to crowded public spaces, such as the New York City subway system, potentially spreading the disease. Researchers hope studying such behaviors will allow officials to implement effective policies. Clari Massimiliano/Shutterstock

NSF Project Gathers Hyper-Localized COVID Data

October 1, 2020
By George I. Seffers
E-mail About the Author

Behaviors near medical centers may inform policies.


New York University researchers are studying the behavior of people leaving healthcare facilities and how they physically interact with the environment—what they touch and for how long, for example. The research will allow the development of localized disease transmission models that can be applied to larger areas, such as entire cities. Potential models could be critical for predicting the continued spread of COVID-19 as well as future pandemics and other disasters, such as chemical spills.

The one-year project, which is funded by a National Science Foundation (NSF) grant of less than $100,000, is known as Developing Epidemiology mechanisms in Three-dimensions to Enhance Response (DETER). The project recruits students to observe human behavior outside of medical facilities, such as hospitals and urgent care centers. In some cases, the students may follow the human subjects as they walk about in public spaces, including nearby subway stations, so that they can learn more about how diseases, including COVID-19, might spread.

The researchers emphasize that the data is entirely anonymous. The students do not take photographs or videos. They do not ask for names or note physical descriptions. They don’t specify male or female, or doctor, nurse or patient. The research is assessed by a review board to ensure it follows ethical and legal guidelines.

Debra Laefer, professor of urban informatics and director of the Citizen Science program within the Center for Urban Science and Progress and the Department of Civil and Urban Engineering, New York University, says it is unusual for researchers to collect and analyze such data during a pandemic. “I think what’s interesting is that nobody has done something like this before—to be able to get this hyper-local behavior data of what people touch and where they went, what they did.”

Laefer serves as the DETER principle investigator, along with co-principle investigator Thomas Kirchner, assistant professor of social behavioral sciences, New York University. Kirchner says the information could help inform policies, such as the shut-down or so-called pause policies seen in New York and many other areas across the country.

“As the policy implementation fluctuates, we expect that transmission rates and other indicators that we can study are also going to fluctuate, and we can help cities optimize the way they use these sorts of policies to be able to make those smarter,” Kirchner says. Moving forward, he adds, the mechanistic approach—the study of the interaction among people and between people and places—can “help us understand how to implement pause-type policies for mitigating transmission of disease.”

The project officially kicked off in March, and researchers have found some of the initial observations surprising. For instance, many people leave a medical facility only to return shortly afterward. In some cases, they leave and return wearing the same personal protective equipment (PPE).

“We saw about 16 or 17 percent of the people coming out of these facilities go back in. People come out, they have a cigarette, they use their phone, they get some food, they go back in,” Laefer reports. “That was something we hadn’t really thought about. You have people coming out in PPE, and they go back in with the same PPE. They’re touching things. They may be tracking the virus to other places … and they may bring stuff back into the hospital.”

In one instance, researchers saw the potential for disease to be spread from a discarded cigarette. “We had a situation where somebody came out of one of these facilities. They were smoking. They dropped the cigarette, and a homeless person then picked up that cigarette,” Laefer adds. “You can’t get this from cellphone data or other forms of large-scale monitoring.”

They were also surprised by the number of items or surfaces people unwittingly touch, including handrails, doors and trash cans. “We found that some of the facilities had no automatic doors, thereby forcing everyone to touch the door on the way in and out. There were other cases where people touched doors that were automatic,” Laefer says.

Also, as New York’s pause order continued, people were more likely to touch more things. “But it’s not so simple. It increases at different rates per facility,” Laefer elaborates.

Furthermore, people touch more items during certain times. “It increases at different rates per day. There are different days of the week where people tend to touch things more,” she adds.

Asked which days those might be, Laefer says it is too soon to discuss the details. “We’re not ready to release that, but there’s definitely a pattern. So, what happens is that we’ll look at one piece of the data for one facility, and now we need to go back and analyze the other 15 facilities and see if this is something that happens everywhere, or only in the Bronx or only at urgent cares,” she explains. “As we start getting into this really nuanced look at the data, we’re definitely seeing some things that people have not talked about or considered before.”

Thomas emphasized that the research is exploratory. “This isn’t a representative sample necessarily of all touching, so it would be a little bit of a confirmatory leap so early in the process to say that touching on Tuesdays is more or something like that,” he says. “But we wouldn’t have thought to look there before. And the idea that we can capture non-random variations in these behaviors is kind of the ‘Hello, World’ in the sense that we can start to measure and understand these behaviors.”

Perhaps not so surprisingly, some people are more willing to wear PPE than others. “What we’re seeing already from some initial data analysis we’ve done is that women wear it more. It doesn’t matter if it’s masks or gloves. It’s everything,” Laefer states. “There is definitely a statistically significant difference between men and women.”

That information could prove useful for public service advertisements. “If you’re trying to do public health announcements, maybe you should be targeting ads to males,” Laefer suggests.

Kirchner points out that there are some stigmas attached to wearing PPE. “I think a challenge for everybody working in this area is the intersection between the science and the social norms, the stigmas and so forth that are developing. We can’t really speak to that side of the equation, but it’s also wise not to be just completely blind and oblivious to it,” he says. “For this sort of work to move forward in a healthy way, we typically would begin to think about stakeholders and community perspectives.”

While the data gathered is hyper-localized in public areas near medical facilities, it can be applied to larger areas, which is one reason NSF officials decided to fund the project, says Scott Freundschuh, program director, Human-Environment and Geographical Sciences, and co-chair, Coastlines and People program, NSF. “They’re collecting data of very hyper-localized behavior, and then we can extrapolate that behavior to larger spaces—neighborhoods and cities—to really understand human behavior and how people are interacting with their environment.”

While the DETER project is focused on the data, that data could potentially be used for 3D models to help illustrate the spread of the disease. “Think about just using a map. You can look at a vector where people go or the movement. But imagine if you can create an environment where you can immerse yourself in that space, and you can walk that trail and see the various things that people touched. You get a better sense of why people are reacting the way they are,” says Arthur Lupia, who heads NSF’s Directorate for Social, Behavioral, and Economic Sciences.

Laefer has used a laser device to scan some of the public areas the human subjects visit, including the New York City subway system. Ultimately, they hope to integrate the 2D records—who touched what when and for how long—with the 3D models. “You would be able to show that in 3D, and then you would have somebody actually kind of walk through this environment,” Laefer says.

Those models could offer insights into what areas might need to be cleaned more often or where officials might need to post warning signs advising people not to touch. Or, in some cases, they could lead to physical changes at some facilities. “For one of the hospitals that we monitored, every person who came out the door touched the door because it wasn’t an automatic door. There were hundreds of people that came out of that door every day. So, that would be an easy opportunity to find those high-touch areas and change policy, or in this case, change the building to retrofit things like that,” Laefer offers.

Kirchner notes that the data will be available to others who can build their own models using free, open source tools such as OpenStreetMap or Mapillary. “Some of these tools can be rolled out and implemented by anyone,” he says. “There’s a huge network, for instance, for humanitarian disaster response. OpenStreeMap is a vital tool to allow low-income areas to respond to earthquakes and so forth using things like cameras and pictures and to populate maps of places that don’t have really good imagery.”

In addition to the present and future pandemics, such 3D models could potentially be used in other disasters as well. “You can think about places that store dangerous chemicals and what can happen if there’s a leak or things of that nature. You still have this idea that particles get on your clothes and that you can move them around,” Lupia says. “Part of the reason for giving the grant is that it’s applicable now, but the depth in which they’re looking at the problem does give the potential to be applicable in the next pandemic, the next accidental release of a chemical and similar projects.”

Enjoyed this article? SUBSCRIBE NOW to keep the content flowing.


Departments: 

Share Your Thoughts: