COVID-19 has dramatically changed our lives and affected how we move, shop, work, and interact with each other. In this dashboard we provide a set of indicators that summarize how people's behaviors adapted in response to the ongoing outbreak.
For more coverage about the MOBS lab’s COVID-19 research, see https://mobs-lab.org/2019ncov.html and https://covid19.gleamproject.org/
Last updated on September 27th.
Report 1 (March 31) Report 2 (May 11)Combined measure, aggregating typical mobility and contact patterns. Percentages refer to the level of the measures with respect to their baseline values in January and February.
To measure contact opportunities, we split each day into 5-minute time bins and we check when two users are co-located within the same geohash. We compute two daily statistics that are proxies, respectively, for the total amount of time spent in contact with other users (contacts duration) and for the distinct number of users encountered within a day in a given location (unique contacts). In the following, index is proportional to the value of these statistics while typical behavior shows how each day compares to baseline values of the statistic in the same weekday from January and February (e.g. the value on Tuesday April 7th is compared to a typical Tuesday in January/February). This analysis explicitly excludes users’ personal areas (i.e. home and work locations).
As a proxy for individual mobility, we calculate the radius of gyration of anonymous users throughout a day. The radius of gyration measures the mean square deviation of the distances traveled by a user on a given day, as measured from the center of mass of the trajectory. Larger radii of gyration correspond to trajectories with positions that are far away from the average position. We compute a daily statistic that is proportional to the radius of gyration for the users living in the selected location. In the following, index is proportional to the value of this statistic while typical behavior shows how each day compares to baseline values of the statistic in the same weekday from January and February (e.g. the value on Tuesday April 7th is compared to a typical Tuesday in January/February).
To measure commuting behavior, we count the daily number of active commuters. A user is defined as an active commuter if within the span of one day they move between their two personal areas. We compute a daily statistic that is the fraction of active commuters relative to the population of the selected location. In the following, index is proportional to the value of this statistic while typical behavior shows how each day compares to baseline values of the statistic in the same weekday from January and February (e.g. the value on Tuesday April 7th is compared to a typical Tuesday in January/February).
The data used in this dashboard are provided by Cuebiq, a location intelligence and measurement platform. Through its Data for Good program [1], Cuebiq provides access to aggregated and privacy-enhanced mobility data for academic research and humanitarian initiatives. These first-party data are collected from anonymized users who have opted in to provide access to their GPS location data anonymously, through a GDPR-compliant framework. In addition to anonymizing all data, the data employs privacy-preserving techniques (referred to as “upleveling”) to reduce risk of re-identification, such as aggregating the inferred home and work areas to the census block group level [2]. This allows for demographic analysis while obfuscating the true home location of anonymous users and preventing misuse of data.
This dashboard includes data aggregated at various levels. We report trends at the national level, state level, as well as the 50 most populous Combined Statistical Areas (CSAs), which can be thought of as large metropolitan areas (e.g. the New York City CSA extends into New Jersey, Connecticut and Pennsylvania)
Disclaimer: The results are estimates. The dataset used here contains a sample of people in the United States, and as such there is bound to be variability in the coverage and representativeness of the results above. The presented material is subject to change as more data become available. Future decisions on when and for how long to relax mitigation policies will be informed by ongoing surveillance. Additional reports are required to assess the level and effectiveness of additional non-pharmaceuticals interventions required to lift current social distancing measures.
* Results for Vermont are currently not reported due to lack of sufficient data coverage.
We thank Ciro Cattuto, Michele Tizzoni, and Zachary Cohen for their help understanding the details of Cuebiq data and Esteban Moro for his comments. We also thank Chia-Hung Yang for coding assistance. MC and AV acknowledge support from Google Cloud Healthcare and Life Sciences Solutions via GCP research credits program. The findings and conclusions in this study are those of the authors and do not necessarily represent the official position of the funding agencies, the National Institutes of Health or U.S. Department of Health and Human Services. BK acknowledges support from the National Defense Science & Engineering Graduate Fellowship (NDSEG). TER, LT, and TL were supported in part by NSF IIS-1741197, Combat Capabilities Development Command Army Research Laboratory under Cooperative Agreement Number W911NF-13-2-0045, and Under Secretary of Defense for Research and Engineering under Air Force Contract No. FA8702-15- D-0001.
Web Development & Design
We thank Agastya Mondal, Robel Kassa, and Nicole Samay for the development of this dashboard.