COVID-19 Mobility Data

Wed, 25 Nov 2020 00:00:00 +0000

We are in the middle of a mind-boggling natural experiment here in the United States. In spite of the advice from the CDC and dire warnings from our nation’s health care experts, millions of Americans will travel over the long holiday weekend. Although the number of people flying is significantly down from last year, there are still large numbers of Americans on the move. The TSA reported that more than two million people went through airport checkpoints last weekend, and the AAA is forecasting as many as fifty-million people may travel.

No matter what the outcome, it is a pretty safe bet that the mobility data collected this weekend will be studied by epidemiologists and public health experts for years to come. In addition to the anecdotal reports linking travel to increased COVID-19 transmission a number of studies including this recent PNAS Report which suggests a “positive relationship between mobility inflow and the number infections”, and this Lancet Correspondence which concludes that a “concomitant increases in mobility will be correlated with an increased numbers cases”, the experts are just beginning to understand the dynamics of mobility and the spread of infection. (See, for example, this Nature paper that claims a “relatively simple SEIR model” informed by the hourly movements of 98 million people can “accurately fit the real case trajectory”).

Acquiring mobility data requires access to large scale infrastructure. Fortunately, several sites are providing access to large scale data sets. The COVIDcast site from the Delphi group provides both R and Python APIs to access the SafeGraph Mobility Data. Click here to see a time-lapse animation of “away from home” data that shows how the cycles of travel vary from before the pandemic up through the middle of this month.

Click here

for another classy dashboard from the University Maryland and the Maryland Transportation Institute that shows how mobility data tracks with COVID cases.

To get your hands on some mobility data in addition to what is available with the Delphi API, try out the covid19mobility package which scrapes mobility data from Google and Apple and look here for the data and R code behind PNAS report mentioned above.

For an in depth look at the issues relating to mobility data and the COVID-19 pandemic, please sign up for the next COVID-19 Data Forum event which will be held at 9 AM Pacific Time on Thursday, December 10th.

Chris Volinsky, Associate vice-president of Big Data Research at ATT Labs will moderate presentations and a panel discussion with Caroline Buckee, Associate Professor of Epidemiology and Associate Director of the Center for Communicable Disease Dynamics at the Harvard T.H. Chan School of Public Health, Dr. Andrew Schoeder, Vice-president Research & Analytics for Direct Relief, and Christophe Fraser, Professor of Pathogen Dynamics at University of Oxford and Senior Group Leader at Big Data Institute, Oxford University, UK.

Finally, wherever your are: please assess the risks of travel for yourself, for your family, and for anyone with whom you may share the air. Stay safe!

More Select COVID-19 Resources

Wed, 03 Jun 2020 00:00:00 +0000

We are over five months into this pandemic, and it is pretty clear that almost everyone is really tired of hearing about it. I myself am totally zoomed out and have already seen too many dashboards. Nevertheless, we are in this for the long run. So from time-to-time, I think it worthwhile to continue to look for tools that can help us make some sense of the continuing stream of incoming data.

First, I would like to draw your attention to the Covid Trends animated dashboard from Physics teacher Aatish Bhatia. The epidemiologists are the experts in this domain, but it is just like a physicist to deliver on insight.

What’s unique about this dashboard is how it beautifully illustrates the consequence of exponential growth. Notice that there is no time axis on the graph. Total confirmed cases are plotted on the x axis, and new confirmed cases in the past week are plotted on the y axis. In this setup, doubling times are represented as straight lines. As you run the animation, time passes and you observe the various countries hugging the seven day doubling time line and then dropping down as the whatever counter measures they are taking get the epidemic under control. This plot makes it clear that while things are opening up in the U.S. we do not quite have the disease under control. Please do watch the short video explaining the graph.

Next, please have a look at the COVID-19 Data Hub, an open source project started by Finance Ph.D. student Emanuele Guidotti with initial IVADO arranged by David Ardia that may very well become the main repository for epidemiologists working with COVID-19 case data. Currently over sixty data sets are available.

All data sets are in a standardized format, and are well documented. Additionally, the site provides R, Python, MatLab, Julia, Node.js, Scala and Excel code to access the data. This project is an extraordinary effort that deserves community support.

Finally for today, I recommend the video recording from the first COVID-19 Data Forum webinar held on May 14, 2020. After the opening remarks by Michael Kane, Assistant Professor, Department of Biostatistics, Yale University which begin at one minute and forty seconds (1:40) into the video, there are four talks, each approximately fifteen minutes long.

The first talk: Modeling COVID19 spread and control: Data needs and challenges by Alison Hill of the Department of Organismic & Evolutionary Biology, Harvard University begins at (5:33). The second talk: Collecting and Visualizing COVID-19 Case Count Data from Multiple Open Sources by independent consultant Ryan Hafen begins at (21:26). The third talk: Spatial and Space-Time Data on COVID-19 by Orhun Aydin of esri and the Environmental Systems Research Institute University of Southern California begins at (38:49). The final talk by Noam Ross of the EcoHealth Alliance and rOpenSci which begins at (56:23), focuses on the genomic data that enables scientists to study the emergence of new diseases.

Enjoy the videos.

COVID-19 Data Forum on R Views

COVID-19 Mobility Data

More Select COVID-19 Resources