Researchers unveil virus surge data dashboard
Northwestern-led group aims to spot infections faster
A Northwestern University-based research team has unveiled an online COVID-19 data dashboard that aims to show more quickly where infections are surging in states and across theworld.
The dashboard joins scores of others related to the pandemic that are hosted by governments, news media, nonprofits and universities. This one crunches the data in a new way that its creators say can flag surges faster and more precisely, before they become overwhelming.
“Basically the whole idea for this is like an early warning system. What we would hope to do with this is to … be able to see when things are starting to stoke and an outbreak is occurring,” said Lori Post, co-lead investigator and director of the Buehler Center for Health Policy and Economics at Northwestern University Feinberg School of
Medicine.
“However, it’s really a difficult time just because all U.S. states are just like out of control right now in America,” said Post, a professor of emergency medicine and medical social sciences.
Unveiled Thursday morning, the dashboard’s underlying programming code collects case data from other dashboards around the world, then converts it into metrics that Post says will provide an early signal of problems in roughly 200 countries.
Some of the metrics, albeit named differently, may be familiar to those who follow COVID-19 data. For example, the new dashboard shows the “speed” of the infection spread, which is basically the seven-day average rate of new confirmed cases, per 100,000 residents.
Other metrics offer anew take on the data. Acceleration looks at changes in speed, with another metric — called “jerk” — looking at howmuch that acceleration (or deceleration) is changing. They’re based on rates computed per 100,000 residents, so different areas can be fairly compared.
Consider the firstweekof November. The model showed Illinois with a “speed” rate of roughly 61 new cases a day, averaged over aweek, but a relatively high acceleration rate that averaged four more new cases each day than the day before. And the jerk rate was a positive number, meaning the rate of acceleration was itself growing fromweek toweek.
In essence, the model reflected the exponential growth in the middle of a massive fall surge of cases in Illinois.
The model’s initial release comes the same day some of its older data, through mid-November, is being published in the Journal of Medical Internet Research. The website includes more recent data for the United States, and Post said the model will eventually be updated daily.
The dashboard reflects how Illinois has recently seen its curve of average daily case growth begin to bend and drop, but public health advocates alsoworry that trend could be short
lived. No one’s sure howthe Thanksgiving break will affect the numbers, with large holiday gatherings and college students returning home.
Post, for one, expects a new surge that will be captured in results that will come in later this month. She said her model has an additional metric that can help estimate the “echo effect” of new infections tied to cases from seven days earlier.
The dashboard joins other, more established forecasting done across the country, and more long
established epidemiological methods to measure infection spread, such as the R factor. Illinois has commissioned other groups of researchers to provide their own weekly updates that, based on these methods, predict how cases, hospitalizations and deaths may change, including ones that forecast serious ICU bed shortages last month if Illinois didn’t tame the infection’s spread.
Post said her group’s dashboard, produced with co-lead investigator James Oehmke and other re
searchers from Northwestern and the University of Florida, is meant to supplementthatworkby offering a farwider, albeitmoresuperficial, look at case data. Their hope is to eventually offer more localized reviews so policymakers can better spot problem areas and more precisely impose restrictions to curb the infection spread.
“We want to make this really usable, so you don’t have to have a Ph.D. in economics to be able to do it,” she said.