Pandemic data mining underscores importance of social distancing in China — and in Seattle

"Stay at Home" sign
A traffic advisory on Interstate 5 emphasizes the “Stay Home” message. (WSDOT Webcam)

Masses of location data gathered in China show that intensive testing, tracking and restrictions on mobility are effective strategies for fighting a coronavirus outbreak. They also show that the strategies need time to work.

And although it’s still early, similar evidence suggests that the strategies are working in Seattle as well.

The evidence from China is laid out in a paper published online today by the journal Science. An international research team, led by the University of Oxford’s Moritz Kraemer, correlated mobile location data from China’s Baidu social media venture with epidemiological data from the Open COVID-19 Data Working Group.

Kraemer and his colleagues said China’s travel restrictions drastically reduced the number of cases that were traced to the outbreak’s epicenter in Hubei province from other parts of the country. Before Jan. 31, there were 515 such cases. After Jan. 31, there were just 39.

Unfortunately, the virus apparently spread through asymptomatic or “stealth” transmission for up to 14 days before it was detected in new locales.

“This is where a full package of measures, including mobility restrictions, testing, tracing and isolating need to work together to mitigate local spread,” Kraemer said in a news release.

Northeastern University’s Samuel Scarpino said political as well as scientific factors have to come into play.

Coronavirus Live Updates: The latest COVID-19 developments in Seattle and the world of tech

“The political will in many countries is lagging behind the spread of COVID-19,” Scarpino said. “Lockdowns and quarantines work, but it takes time, because the infection rate is five to 14 days ahead of any mitigation action taken.”

In the United States, the initial response to the outbreak was hampered by the lack of adequate testing for the virus. But over the past few weeks, an increasing number of states — including Washington state — have gotten up to speed on mobility restrictions, and are ramping up tests as well.

In the Seattle area, officials made a series of moves to tighten up on social interactions, starting with a work-from-home recommendation from Seattle-King County Public Health on March 4 and building up to Washington Gov. Jay Inslee’s stay-at-home order on Monday.

King County spokesman Norm Mah told GeekWire that it’s still too early to assess the impact of those moves.

“The effects of social distancing measures occur after approximately two weeks, because of the incubation period of the infection,” he said in an email. “Indicators can include the rate of COVID-19 case reports, hospitalizations and deaths, impact on the health care system and measurements of social distancing itself such as commuter traffic, ridership on public transit, and number of people teleworking.”

Mah said the number of confirmed cases isn’t a good indicator for assessing the spread of the disease, because an increase in cases could reflect greater availability of testing rather than wider transmission of the virus. “Once there is stable and more widespread testing capacity, our measurements will be more reliable,” he said.

Other indicators provide some cause for hope. Let’s start with the mobility data.

King County residents have cut their travel distances dramatically over the first weeks of March. In this data by Descartes Labs, a normal level of movement is indexed as 100, and anything below that is a decline. For example, if people travel 60% less than normal, the index would be 40. The distance is calculated as the median of the maximum distance traveled away from the starting point of the mobile device. (Descartes Labs Data; GeekWire Image)

Descartes Labs, a geospatial analysis company based in New Mexico, used data collected from mobile devices to determine the normal amount of travel in a given area, and then looked at changes in mobility linked to the coronavirus outbreak.

During the first three weeks of March, when mobility restrictions were being put into effect, Descarte Labs found that King County residents reduced their travel by roughly 80%.

Another study by Citymapper, which makes a mapping app for smartphones, analyzed queries from the app’s users to calculate a mobility index for cities. That analysis showed a steady decline in travel by Seattle-area users this month, with the reduction in movement falling by 90% as of Tuesday.

Using data from its mapping app, Citymapper estimated the change in travel during the novel coronavirus outbreak. (Citymapper Image)

Experts caution that such data sets have significant limitations. That’s particularly true when trying to understand whether people are following the rules about social distancing — namely, staying 6 feet or more away from those not in their immediate family or household.

“Social distancing and distance traveled aren’t the same thing,” said Tim Althoff, assistant professor with the University of Washington’s Paul G. Allen School of Computer Science & Engineering.

To get more meaningful data about how people are behaving would require tighter monitoring of people’s interactions with each other, but Althoff said that touches on issues of personal privacy and consent for the use of sensitive data.

In the absence of widespread testing for the virus, is it possible to correlate location with the prevalence of COVID-19? A California-based biotech venture called Kinsa Health is trying to do that more than a million smart thermometers.

Kinsa has been analyzing data gleaned from users’ smartphone apps to identify areas where the incidence of fever is higher than historical levels — and overlaying those atypical fever levels on county-by-county maps.

The maps don’t track COVID-19 infections directly, but they can point to places where viruses are causing more than their usual share of fevers.

Kinsa fever chart for King County
A chart of fever incidence in King County shows an atypical spike starting on March 3, rising to a peak on March 9, and fading away to more typical levels in the March 17-24 timeframe. (U.S. Health Weather Map / Kinsa Health / Oregon State University)

When Kinsa looked at the data from King County and surrounding areas, it found that the fever incidence rose to atypical levels around the start of the month, peaked on March 9, and gradually declined to expected levels in the past week.

If you assume that it takes someone with COVID-19 to develop symptoms, you could arguably correlate that rise and fall with the significant spread of the virus in late February, and the onset of social-distancing behaviors in early March.

“Because of social distancing, we would expect the levels to drop, which is what you’re seeing,” Kinsa spokeswoman Nita Nehru told GeekWire. “Every day that we see the numbers drop is an indication that social distancing is working.”

Previously: Seattle lags SF and NYC in getting people to stay home

Nehru cautioned, however, that the picture is more complex than it looks. When people avoid each other, that cuts down on the communicability of colds and flu as well as coronavirus.

“This decrease is going to be presumably across the board,” she said.

This season, some of those seemingly typical fever cases are almost certainly being caused by coronavirus rather than flu or cold virus — and that underscores the fact that we still have a long way to go to control COVID-19’s spread. Like the researchers who studied China’s outbreak, Kinsa’s fever-watchers say it could take several more weeks of social distancing to tell the tale.

For Nehru, social distancing in King County isn’t merely a statistical curiosity. It’s personal: She grew up in Redmond and stays in regular touch with her mother, who’s in her 60s and works as a weight-loss adviser in Bellevue.

“I definitely want her to stay home,” Nehru said.

David Pigott of the University of Washington’s Institute for Health Metrics and Evaluation is among the co-authors of the paper published by Science, “The Effect of Human Mobility and Control Measures on the COVID-19 Epidemic in China.” Kinsa Health created the U.S. Health Weather Map in collaboration with Oregon State University’s Benjamin Dalziel.

View original article here Source