Facebook using artificial intelligence to forecast COVID-19 spread in every U.S. county

San Diego projected to hit 78,500 cases by Nov. 30
Posted at 3:28 PM, Nov 20, 2020
and last updated 2020-11-20 21:26:04-05

SAN DIEGO (KGTV) — State officials hope California’s new 10 p.m. stay-at-home order will slow the spread of COVID-19, otherwise, another 10,000 San Diegans are projected to contract the virus in the next 10 days.

That’s according to a new county-by-county forecast from Facebook, which rolled out the prediction software last month.

Facebook projects L.A. County will see the second-largest increase in cases in the country by November 30. San Diego County is projected to add the 15th most cases, reaching a total of 78,594 infections by Nov. 30.

The two-week forecast was released before Governor Gavin Newsom announced enhanced restrictions. Facebook will release a new two-week forecast next week.

“Many other forecasts around the world are only predicting caseload at a country or state level,” said Laura McGorman with Facebook’s Data for Good team. “We’re trying to be much more local in this approach because we know so much of the COVID-19 response is in fact local.”

McGorman said the forecast tool could help county and state officials plan hospital bed space, ICU capacity, ventilators and other critical supplies.

The map is powered by artificial intelligence that draws on seven kinds of data. They use outside metrics like confirmed cases, doctor visits and the weather combined with information Facebook collects like a survey of people’s symptoms and GPS location data.

The location data helps gauge whether people are staying home and isolating or circulating among the community, according to McGorman.

Facebook only pulls data from users who opt in and no, they’re not reading your posts. All of the information, which is aggregated to the county or state level to protect privacy, is available for the public to download.

The tech giant initially started Data for Good three years ago to help with disaster-relief projects, McGorman said. Among other applications, their location data can quickly predict if evacuations are working.

“Typically, groups like the Red Cross have to knock door-to-door to see if people are still home, or wait for people to show up at shelters to see if they’ve gotten out of harm’s way,” she said.

The software has also been used to detect network outages when a hurricane knocks out cellphone towers, she said.

Facebook’s COVID-19 mapping tools have been used to inform policymakers in New York and Mexico, and the data has helped analyze the effectiveness of stay-at-home orders in California, McGorman said.