Experts explore digital contact tracing for COVID-19
ASU geospatial experts weigh in on technology, privacy and a path forward
In an effort to halt the spread of coronavirus, more countries are exploring the use of a wide range of technologies for the purpose of digital contact tracing, that is, leveraging personal data to identify who may have been exposed to someone with the disease.
However, as decision-makers and health professionals consider how technology can be used to protect public health and minimize social and economic impacts, there are many things to consider.
Arizona State University’s School of Geographical Sciences and Urban Planning and Spatial Analysis Research Center (SPARC) brought together geospatial experts from across the nation in an online conversation about both the technical and ethical issues of digital contact tracing in response to COVID-19.
Titled “Digital Contact Tracing and Surveillance: A National Conversation with Geospatial Experts,” the conversation answered questions about the accuracy of cell phone GPS data, how social media can be used for tracking and looming privacy issues.
“When you have a pandemic — a contagion and spread, spatially and temporally constrained — geospatial data scientists are experienced in understanding and analyzing things that have dynamics through space and time,” said Trisalyn Nelson, director of the School of Geographical Sciences, professor of geospatial data science and organizer of the panel.
“It’s time that the geospatial community stood up and provided solutions quickly to then really press the issue.”
Contact tracing technology around the globe
Song Gao, assistant professor in the Geospatial Data Science Lab at the University of Wisconsin, Madison, said many governments in East Asia have automated their contact tracing approach by leveraging multiple technologies in conjunction with one other, integrating mobile phones, geographic information systems (GIS) and health informatics.
GPS in mobile phone apps has been used to track an individual’s trajectory and the places they’ve been to, while Bluetooth has been leveraged to identify people an infected person may have had close contact with. In China, the government also uses QR codes that link individuals' health record information. People are required to scan the QR code if they want to enter a public space such as getting on public transportation or staying in a hotel.
“This process builds a database of all human-to-human contact networks and human-place interactions,” Gao said. “By doing it, if a person is infected, you can track the people that may have been there at the same place and time, and authorities can quickly contact and notify them.”
The purpose of 'trace and alert'
While the benefit use of different technologies can vary for different tracing objectives, for the purposes of leveraging technology to identify people who may have been in contact with someone with the disease, panelists said Bluetooth provides the most anonymity and tracking accuracy.
“Where we want to be is having some system that basically will help us identify someone who has been in contact with somebody with the virus,” said Stewart Fotheringham, Regents Professor in the School of Geographical Sciences and Urban Planning and director of SPARC.
“We want to have a system where everyone is carrying their device that just recognizes and sets a log who you’ve been in contact with within two to three meters, and we have that technology; it’s called Bluetooth.”
Fotheringham and others contend that because Bluetooth does not track a person’s trajectory, it avoids privacy concerns related to locating people in a specific space. For example, Bluetooth can tell you if you had been in proximity with someone else who has identified themselves as disease positive, but alone can not tell you whether you came into proximity with the infected person while you were at the post office, the park or the grocery store.
Peter Kedron, assistant professor with the School of Geographical Sciences and Urban Planning and specialist in spatial and economic analysis, said that is a key piece of the privacy conversation.
The challenge to privacy: Reidentification
Kedron cautioned that with individual trajectory data — which is collected by things like GPS and cell towers — it is becoming easier to identify individuals from anonymized data sets. Research has shown that it is possible to reidentify individuals even when anonymized and aggregated data sets are incomplete.
Kedron pointed to two studies, one that suggests that if you have a person’s home location at a census block group level and you have their work location at a census block group level, you can potentially identify 50% of the U.S. population down to 1 in 10 people. Additionally, other research suggests that having as few as four to five points of high resolution spatial temporal data is enough to identify more than 90% of individuals.
A partial way around this is for solutions that focus on using relative space as opposed to absolute space, similar to what Apple and Google systems are proposing to do — using Bluetooth and having encrypted lists of who has been in contact with whom.
“We want to seriously consider this trade-off between sharing location data publically versus some privacy concerns because again reidentification is probably likely or possible in a lot of cases, and then in just a larger sense, we want to try to maintain and preserve notice and consent,” Kedron said.
Centrally stored data
Panelists also raised concerns about the importance of identifying who would own any data collected and where it would be stored.
“For privacy, I think it’s important that whatever location information is collected, it has to stay on the individual’s device and only when that individual has been diagnosed positive, then provide that location information to a public health server for example,” said May Yuan, professor of Geospatial Information Sciences at the University of Texas at Dallas.
“If you look at it, the public health server really does not need to have an individual’s trajectory, they only need what places this individual has visited in the last 14 days and how long that the individual has been at those places and what time.”
A lot of data storage and use concerns will depend on who manages the contact tracing system, the panelists said. Many believe that if independent companies run the contact tracing technologies there won’t be many incentives to protect our individual privacy.
“If we are going to task an organization in the U.S. to be responsible to manage this kind of system, which the American public is going to trust, I think the CDC has to take the lead with organizing the database, then people will be willing to share,” Yuan said.
Ming Tsou, professor and director with the Center for Human Dynamics in the Mobile Age at San Diego State University, agreed, saying that of all the organizations that the American public would trust the most, similar to how the U.S. Census government organization collects and stores data, in this instance the CDC is best to take the lead.
The future of privacy
Before any contact tracing system is implemented, panelists agreed that there needs to be more public dialogue about what contact tracing could mean for the future of personal data collection and privacy.
“We need to design this system that puts a boundary or a buffer in what we allow that data to be used for and how long we keep it,” Fotheringham said. “I think it’s important that we don’t use this time during a pandemic to make this thing (digital contact tracing) a norm because it may not necessarily be a norm that we want, and we should have a conversation about that publicly.”
Amy Frazier, assistant professor in the School of Geographical Sciences and Urban Planning, agreed, saying that before any changes are made, we must remember what our baseline of privacy and personal data access is today.
“If we move forward and we don’t think about what our starting baseline was and what we were comfortable with, it becomes very hard to go back once things are safe and we don’t need to do contact tracing anymore.”
Kedron added, “We’re not just making decisions for now, because those decisions tend to be hard to draw back once the systems are out into the world, and we need to have a conversation about what those things are going to be.
“We want to give clear notice to people, we want to let people know what happens to their data, how it’s stored, we want to notify people when things change, and I think most importantly, in this moment of exception and concern, we want to have clear plans about how this system is going to sunset and how this is going to be ended as the pandemic slows and overall,” Kedron said.
ASU geospatial response
The panel is just one of the efforts the School of Geographical Sciences and Urban Planning is undertaking to pivot quickly and leverage the school’s expertise for supporting COVID-19 response.
Nelson says having the expertise of ASU’s SPARC enables this expedited action.
“Because we have SPARC at ASU, we are able to pull together, very quickly, these big national conversations around critical issues,” Nelson said. “We can focus on COVID-19 because people are asking and because we are leaders in the field of geospatial science. We are good conveners for these types of conversations throughout COVID and into the future.”
The School of Geographical Sciences and Urban Planning has put out a call for any individual or organization that needs help to mobilize the department’s skills in GIS to help fight COVID-19. Students and faculty can map, analyze and create dashboards with data.
Additional panelists included Michael Goodchild, research professor, Spatial Analysis Research Center, Arizona State University; Yingjie Hu, assistant professor, National Center for Geographic Information and Analysis, University of Buffalo; Bo Zhao, assistant professor, Department of Geography, University of Washington; and WenWen Li, associate professor, School of Geographical Sciences, Arizona State University.