A team of epidemiologists and computer scientists believe the same kind of computational forecasting that predicts the spread of a virus can also help guide the spread of equity and social health.
November 18, 2020
Right now, a start-up software company that spun out of the Graduate School of Public Health at the University of Pittsburgh is working with school districts and public health agencies to help make critical decisions about how to operate during the COVID-19 pandemic. The company has developed software to predict the progression of disease outbreaks and other viral phenomena.
Their immediate focus is on the pandemic, but the long-term potential expands beyond stopping the spread of microscopic contagions into helping policymakers solve societal challenges like a lack of affordable housing and the spread of poverty. Those problems, too, the founders say, have qualities of contagion.
For school districts and public health departments, the company’s work in this moment relies on selling a subscription service through which it can precisely model how many people within a school or district would likely get infected under all the different operating models leaders are considering. Those could include opening fully, going hybrid, giving sports teams permission to practice, applying mask-wearing procedures, and so on.
As the school year goes on and conditions in communities change (rates of infection rise or fall, for example, or restaurants open or shut down again), districts that subscribe to the service will be able to update the predictive outcomes of their decisions.
The company is called Epistemix, and it’s currently working with districts to model school opening decisions in Rhode Island, Pennsylvania, West Virginia, Indiana, Oklahoma, Texas, and California. John Cordier (San Antonio ’14), who got double master’s degrees in business and public health after finishing the corps, is its CEO. Cordier co-founded the company with its chief technology officer, John Grefenstette, and the former dean of the university’s school of public health, Donald S. Burke.
Burke is an epidemiologist who has been at the leading edge of developing vaccines for epidemics like HIV/AIDS. He’s led efforts around the world on understanding infectious diseases like dengue. For a decade, he and members of the Epistemix team have been working to develop a modeling system to predict the progression of disease outbreak. It’s called FRED (Framework for Reconstructing Epidemiological Dynamics), also named for that Pittsburgh hero, Mr. (Fred) Rogers.
While the simulator was in development, one project its developers took on was modeling the tipping point for measles vaccination rates. Cordier said researchers were able to show that once the vaccination levels within a community dip below a certain threshold (around 92%), suddenly one case of measles can become an epidemic.
The utility of using predictive software during a pandemic is clear. But when you model an epidemic through a computer simulation, it turns out that’s a close analogue for other things that “go viral” and spread through human populations.
Cordier and Burke have been working with graduate students at the public health school in Pittsburgh to determine if there’s a way to use FRED to model what they call “the epidemiology of privilege.” The goal would be to use predictive software to develop policies that apply leverage wherever it’s most effective (say, to community economic development) to make systems, including education systems, equitable.
“We haven’t done that yet,” Burke cautions. Modeling social contagions is a leap from modeling a biological disease, in part because those who want to use the tool would first need to agree on the most important factors that contribute to individuals’ overall well-being, access, and social mobility.
But Cordier is trying to raise awareness that “we have a software tool that can enable subject experts in housing, education, sociology, or anthropology to look at the outcome of policies.” If it works, the hope is it could also reveal the swiftest pathways to a just world.
“One of the other things superintendents are really struggling with is trying to communicate messages to teachers and parents to justify why they’re not opening in full, or staying all remote, or doing a hybrid.”
Better than Political Decision-Making
Explained simply, Epidemix software gives leaders in a pandemic tools to make decisions that aren’t based on politics or the particular mental models of the world held by a few key decision-makers. During the decade in which the founders have been developing the platform, there’s been a movement in science to make critical public health decisions more analytical; to build better computational models of the world than you could derive from a group of leaders sitting around a table arguing over who has the better idea.
Right now, districts don’t lack for sources of information on patterns of disease spread, or how other districts are operating, or best practices for continuing education in a pandemic (including a COVID-19 School Response Dashboard co-created by national associations of principals and superintendents). Some of that information is coming from local and state health departments and some from federal agencies like the Centers for Disease Control. But much of it conflicts, or is incomplete, or not localized.
Epistemix aims to be different by building computational models of places as small as towns of a few hundred people, or even one school. Pulling from multiple, de-identified public data sets (like the Census Bureau’s American Community Survey) to represent geography, demographics, social contacts, environmental interactions, and the predictive responses and behaviors of every person in a given place, it stitches all that data together to make a computational simulation of a place like Pittsburgh. It’s Pittsburgh in a computer.
Then epidemiologists, social scientists and computer scientists combine their data on the social dynamics of how humans behave with the biological model of how diseases spread. From there, they can run any number of scenarios to help a state or district predict what level of infections could arise based on different school operating strategies.
For school district leaders who “face the challenge of how do I choose one operating strategy or opening model over another,” Cordier says, it does more than give them a way to make decisions. It helps them communicate a scientific rationale in a highly politicized environment. “One of the other things superintendents are really struggling with,” Cordier says, “is trying to communicate messages to teachers and parents to justify why they’re not opening in full, or staying all remote, or doing a hybrid.”
Agencies and districts can pay to work with Epistemix or they can work with a third party, like a foundation, to pay for the service.
Modeling Educational Equity
While the Epistemix team is focused at the moment on helping model responses to the pandemic, the team is eager to develop FRED to help guide leaders to make complicated decisions around other social contagions. An example is the opioid epidemic that’s emerged out of a conglomeration of conditions including lost jobs and education gaps and affordability. In recent years, Burke said, “We’ve started to put together courses at the school of public health to teach how to think about these complicated problems in ways that allow you to approach them as a system, and to try to get some numbers around these complicated social dynamics.”
If nothing else, the tool is a forcing mechanism to bring large numbers of people together to decide what really are the most critical factors to achieve equity, and then to test those assumptions over time by running simulations.
“We need to retain a level of humility when we talk about predictional forecasting,” Burke says. “But once you start to see the patterns, and you have what you think are the major interacting components, it should be possible to work through the trends and the likely outcomes. And importantly, you can question your assumptions and test what really matters in a simulation as it plays out over time.”
Students Ask Why
Sarah (Mili) Milianta-Laffin (Houston ’06) has been having hard conversations with anxious students at Ilima Intermediate School. In Hawai’i, where she teaches, the numbers of coronavirus infections were dropping in October. But as the state opened its doors to tourism (with new travel rules requiring tourists to document negative COVID-19 tests administered by state-approved testing facilities), students worried that the flow of people into their communities could bring danger. They questioned whether government leaders were prioritizing tourist dollars over community health.
Milianta-Laffin has been discussing with students how they can advocate for their safety by drawing parallels to activism during the AIDS crisis of the 1980s. Students are exploring how some researchers are searching for a COVID-19 vaccine using the same processes developed to understand HIV and AIDS.
Milianta-Laffin sponsors her school’s Gay Straight Alliance (GSA), the Ilima Rainbow Royales. In May, the Rainbow Royales won the national GLSEN GSA of the Year award. For other educators interested in helping mature, older students draw connections between the AIDS crisis and COVID-19, she recommends:
“Lessons from AIDS for the COVID-19 Pandemic” published on October 1 in Scientific American
How to Survive a Plague, book and documentary film by David France
By Joel Serin-Christ
Cover illustration by Hui Yang