Two weeks after beginning to survey users about disease symptoms in an effort to track the spread of COVID-19, Facebook said Monday that early results showed promise — and that the effort will roll out internationally beginning Wednesday. Carnegie Mellon University, which conducts the survey, said that 150,000 people a day are now submitting symptom reports after clicking on a link in the News Feed. The university does not share any information about symptoms back to Facebook.
In an op-ed in the Washington Post, Facebook CEO Mark Zuckerberg said that initial findings shared by Carnegie Mellon correlate with public data about confirmed cases of COVID-19, an important signal that symptom surveys will soon be able to forecast the disease’s spread. Facebook will use aggregated data from the university to produce interactive maps based on symptom surveys and will update them daily.
“I’ve always believed that helping people come together as a community will help us address our greatest challenges — not just by sharing our experiences and supporting each other in crises but also by working together at scale to solve problems,” Zuckerberg wrote. “The world has faced pandemics before, but this time we have a new superpower: the ability to gather and share data for good. If we use it responsibly, I’m optimistic that data can help the world respond to this health crisis and get us started on the road to recovery.”
Facebook also announced a partnership with the University of Maryland to begin collecting symptom data globally. If Facebook users report symptoms in their own countries at rates similar to their participation in the United States, the data could prove to be an effective map of potential hotspots around the globe, said Ryan Tibshirani, an associate professor at Carnegie Mellon who is helping to lead the school’s partnership with Facebook.
In an interview with The Verge, Zuckerberg said that global maps could serve as a reality check in places where elected officials have been slow to acknowledge the spread of COVID-19 within their borders.
“Some of these governments, frankly, are not excited about the world knowing how many actual cases there might be, or indicators of how it’s spreading in their countries,” Zuckerberg said. “So getting that data out there is very important.”
As the pandemic has spread around the world, Zuckerberg has sought to demonstrate how Facebook’s massive global reach can be used to help elected leaders and public health officials make more informed decisions about their pandemic response. Last month, Facebook began inserting a “COVID-19 Information Center” into the News Feed, showing users vetted information from public health authorities. More than 350 million people have used the center to date, Zuckerberg said.
The moves come as Facebook remains under scrutiny for its practices related to competition, data privacy, and allowing the spread of misinformation. The company is being investigated by the Federal Trade Commission and a coalition of states attorney general over antitrust and privacy issues.
As the pandemic worsens, Facebook has worked to show how aggregated, anonymized data collected at global scale can be a force for good. Earlier this month the company also released mobility maps that use smartphone location to illustrate the degree to which Americans are complying with state stay-at-home orders. (Google has released a similar tool using its own location data.)
At the same time, the company has come under continued criticism for its role in spreading misinformation related to the coronavirus, and more recently for its use in organizing protests against stay-at-home orders. On some level, Facebook’s corporate initiatives to reduce the harms generated by the pandemic are undermined by the behavior of its user base — and at this point it’s unclear which of those will prove more consequential in the end.
In a separate announcement on Monday, Facebook announced additional partnerships with researchers to use machine learning techniques to do additional forecasting. In New York, the company is working with New York University and Cornell University to create forecasts for the spread of the disease at the county level, as well as predictions for how a rise in cases will affect demand for services, ventilators, and personal protective equipment at area hospitals. One approach will use machine learning technology from Facebook to analyze de-identified X-rays and computerized tomography (CT) scans in an effort to predict patient outcomes and equipment demand.
Zuckerberg said that while Facebook would not seek to interpret the symptom data it shares with researchers, its size has enabled it to make a significant contribution to the public health response.
“What we can do is help them get a survey out to a large number of people quickly, and on a daily basis,” he said. “Since we have a basic understanding of who people are, we can make sure that it’s sampled properly. We’re in a relatively unique position where I don’t think that there are that many institutions in the world that could stand up a survey like this — across the country, much less across the world.”