Pedestrians walk in San Francisco's Mission District on April 23, 2020. (Beth LaBerge/KQED)
You may know by now that in order to make, say, a weather app or Google Maps work correctly, your phone tracks your location as you move about the world.
That functionality came in handy for a study about coronavirus transmission, recently published online in Nature. The researchers used anonymized phone data to track and model how people’s excursions have driven the course of the pandemic in the 10 biggest U.S. metropolitan areas. Using a computer model to build a network of where people visited and for how long, the researchers were able to infer which locations were hot spots for virus transmission.
"There is a very small number of places, about, let's say, 10%, where 85% of all infections happen," said Jure Leskovec, a Stanford computer scientist and one of the study's authors. The research, conducted by a group from Northwestern University and the Chan Zuckerberg Biohub as well as Stanford, looked at infections between March 1 and May 2 of this year.
Indoor sit-down restaurants, cafes and gyms were the top three points of transmission, and crowding in any indoor business increased the risk of contracting the virus.
The data also show that grocery stores in low-income areas were more crowded, with longer customer visits.
"We found out that even a trip to a grocery store is twice as risky for an individual from a low-income neighborhood versus an individual from a high-income neighborhood," Leskovec said.
In addition, residents of low-income census tracts reduced their movements less than those of higher-income tracts, the analysis found.
"For example, in Chicago, poorer neighborhoods decreased their mobility 30% less than what richer or higher-income neighborhoods did," Leskovec said.
He partially attributes this to low-wage service jobs requiring in-person work, as opposed to higher-wage positions that can be carried out online.
"One of the big achievements of this study is to highlight the occupational vulnerability that people on the lowest ends of our occupational structure have had to face," said Merlin Chowkwanyun, a public health professor at Columbia University, who was not involved in the research.
He favorably compared the study's ability to specify what was happening on the ground compared to the more generalized data available on COVID-19 dashboards like the one from Johns Hopkins University, which provide a limited picture in terms of cause and effect. Those illustrate a good macro view of the virus's growth rate, but leave public health officials and researchers to read the tea leaves amid the peaks, valleys and plateaus of the data.
"Afterwards we kind of guess, ‘Oh, maybe this event like a mask ordinance or a stay-at-home order or a big rally or whatever, may have caused this or that jump,'" Chowkwyanyun said. "But it's all kind of post hoc speculation.
"What was cool about this was it used some of that data, but it overlaid it with very, very specific data about individual behavior that you could actually document," he said.
The analysis shows that despite the current surge in cases, public health safety precautions are working insofar as they decreased the probability of the virus jumping from one person to another.
"Even though the numbers are rising, they are actually lower than what our model would predict,” said Leskovec. "We attribute that to the fact that people are wearing masks."
Importantly, the study also confirms that capping a business’s peak capacity can reduce transmissions even though the overall number of customers permitted inside is decreased by just a small amount.
In the San Francisco Bay Area, the team calculated, putting a cap of 30% of capacity for the number of people allowed indoors at the same time leads to a 600% decrease in transmissions, while still permitting 70% of visits overall.
"The conclusion here," said Leskovec, "is that you can get quite a large benefit in terms of infections stopped."
The most important takeaway for the public, he said, is simply: "Stay at home. Stay at home."
Failing that, you should avoid dense crowds where people congregate for longer periods of time.
Get the best of KQED’s science coverage in your inbox weekly.