Produced by the Office of Marketing and Communications
System to Use Social Media, GPS, Government Data to Improve Predictive Models, Communications
By Tom Ventsias
Illustration by iStock
Mandated face coverings vs. no masking. Fourteen days of isolation or five. Online schooling or crowded classrooms. The ongoing COVID-19 pandemic at times seems like a cacophony of mixed messaging from public health experts and government officials.
Much of this variance stems from the evolution and tenacity of the virus itself. Yet other factors—pandemic fatigue, physical location, demographics, politics, and the timing and tone of the messaging itself—have fueled varying levels of public skepticism and confusion.
To meet this challenge, University of Maryland researchers are developing sophisticated predictive models and best communication practices needed to combat future pandemics. They’re crunching voluminous data from the current pandemic—analyzing social media content, epidemiological statistics and public statements from officials—to build a seamless, end-to-end network that considers complex and interdependent biological, environmental and human factors.
Their work is funded by a $1 million grant from the National Science Foundation, part of the organization’s new Predictive Intelligence for Pandemic Prevention Program.
A main goal of the UMD project is to develop a digital platform, called PandEval (pandemic evaluation), that can zero in on specific locales, offering a level of detail not widely available during the current pandemic.
“What we’ve seen is a need to improve messaging and policymaking at the local scale,” said Neil Sehgal, an assistant professor of health policy and management in the School of Public Health. “Public acceptance for health-related mandates—things like a statewide shutdown of non-essential businesses—could look very different in Montgomery County than on the Lower Eastern Shore.”
Sehgal, whose work is focused on novel and emerging digital health technologies and their applicability to health care delivery and outcomes, is joined on the project by a multi-institutional team of computational social scientists and data scientists, public health experts, biostatisticians and epidemiologists.
It includes Louiqa Raschid, a dean’s professor of information systems in the Robert H. Smith School of Business who is principal investigator of the award; Vanessa Frias-Martinez, an associate professor in the College of Information Studies; Xiaoli Nan, a professor of communication in the College of Arts and Humanities; Kristina Lerman, a professor of computer science at the University of Southern California; and Eili Klein, M.D., an associate professor of emergency medicine at Johns Hopkins University.
Raschid and Frias-Martinez have joint appointments in the University of Maryland Institute for Advanced Computer Studies, which is providing administrative and technical support for the project.
To develop robust algorithms for the PandEval platform, the researchers are curating data that includes almost two billion Twitter posts since January 2020, social media captures from Facebook, GPS digital footprints from location intelligence companies, face masking statistics from a New York Times database and inoculation data from the Centers for Disease Control and Prevention.
The team will use Twitter and Facebook posts to develop social media-based models of community beliefs and attitudes, offering a window into areas like science skepticism, concern about vaccine safety, a lack of trust in public officials or an unwillingness to contribute to the public good.
Nan and Sehgal are also developing digital tools to evaluate the effectiveness of public health messaging, with a focus on building models that help identify the best person or organization to deliver the right message at the right time.
Frias-Martinez will use her extensive experience in mobility tracking to analyze the GPS data, creating new models to guide safe behavior during pandemics. The software would track activities via smartphones or other mobile devices, and then match it to disease vector models, offering actionable data on whether people should work from home or use public transportation systems.
“We think the benefits of PandEval will be twofold: increasing trust and confidence in our public health infrastructure and giving decision makers epidemiological models that are customized to specific population segments,” said Raschid. “This can be invaluable for things like vaccine rollouts and health-related mandates.”
Communication Computer Science Health Policy and Management Information Systems Social Media Research
Maryland Today is produced by the Office of Marketing and Communications for the University of Maryland community on weekdays during the academic year, except for university holidays.
Faculty, staff and students receive the daily Maryland Today e-newsletter. To be added to the subscription list, sign up here:Subscribe