Mapping Localized COVID-19 Risk in the U.S.

Date:
Move slider or click play to see values over time

About the Map

This map was created as part of a toolkit we are using to better understand how our capabilities can be incorporated to the various efforts to fight COVID-19. Both our core computational antibody optimization offering (previously in stealth, but which we are announcing today and making available to any party making anti-COVID-19 antibodies - free of charge) and our population-scale NGS diagnostic protocol have the potential to play valuable roles during various phases of the pandemic. However, given our own extremely limited resources and the broader resource constraints faced by our industry, it is important we have a framework in place for ensuring that we are developing to an achievable scale. It is also important that we understand what criteria should be used to prioritize deployment.

As events have unfolded, the importance of highly-localized phenomena in determining what kinds of capabilities can and should be brought to bear, and how this affects the relative applicability of the solutions we are developing, has been highlighted again and again. The ability to answer questions around such matters as where best to try to deploy our mass-parallel NGS-based diagnostic protocol in the absence of a specialized facility, or whether it is even possible to manufacture enough antibody to be relevant to the outbreak (and how), once a viable antibody or cocktail of antibodies has been “discovered,” is almost entirely reliant upon having a detailed understanding of localized outbreak dynamics. We developed this map, and related tooling, as a means of incorporating such localized phenomena into our decision-making.

We will be elaborating in the coming days on how we are using this map and other tools to work through these and other questions, and we will be both exposing new map layers and making available as much of the underlying calculus as we can as we progress. In the interest of providing a slightly different perspective on the pandemic and its impact on the United States, we are launching the map today with a focus on highlighting the risks local and regional health care systems face as the pandemic spreads across the country.

We have been inspired and humbled by the response to the outbreak, and we produced this resource to address both our own needs for COVID-19 planning, and also in the hope that it can help as many people as possible contextualize, plan, and respond to COVID-19.

We would love any feedback you may have - be it questions, comments, requests for analysis, or even complaints. Whatever the context or content, please do not hesitate to reach out to us. And in the meantime, we hope you and yours remain as unscathed as possible by this outbreak, and that life returns to a healthy normal, soon.

Reading the Map

LOCAL REGIONS AND LAYERS

Data on the map is presented by census statistical areas, as defined below. We have aggregated county-level data from within these regions as we believe it best represents the true picture and risk associated with each region. For portions of the U.S. not covered by an MSA, we present county-level data. We are not currently displaying data for cases that certain states have not yet assigned to a specific location.

The default layer of the map highlights the ratio of the aggregate number of cases (adjusted for deaths) to the number of ICU beds in a region. While not perfect, we believe this is a crude-but-good indicator of the risk a local health system faces (of being overwhelmed by an influx of cases). This is a critical level to monitor in order to prevent very poor outcomes seen in certain other locations.

Additional highlight layers can be displayed clicking the buttons below the map. Currently, we are presenting:

  • Cases per 10,000 People: The total number of cases in a region compared to the 2018 U.S. Census population estimates for that region.
  • New Cases: The number of new cases in a region from the latest daily update to this data.
  • New Cases per 10,000: The number of new cases in a region from the latest daily update to this data, compared to the total population of the region.
  • Deaths: Total number of deaths from COVID-19 as reported by local and state departments of health.
  • Population: The 2018 census estimates of population for a region.
  • Cases: Cumulative number of cases in a region
  • Change in New Cases in the Past Week: Compares the number of new cases (averaged over 7 days, to avoid daily differences) compared to the number of new cases a week ago (averaged over 7 days).
  • Deaths / Case: Provides the ratio of deaths to total cases. This metric does not equate to a mortality rate, since active cases may result in additional deaths or recoveries, but is a relative approximation for comparing regions. This metric is not calculated for areas that have less than 20 cases.

Additional information on each region is displayed when clicking on or moving over a region.

DATA AND SOURCES

Data has been retrieved from the public sources noted below. We claim no ownership of the source data. We have completed various data analysis and transformations to present these maps. All of the data, analysis, and data transformation code can be found in our GitHub repository. You are free to copy, download, fork, or to otherwise use any of our work for your own purposes (with attribution - especially to our sources). We will be maintaining and adding data to the map for as long as the outbreak is active.

We have retrieved source data from the following sources:

DEFINITIONS

In addition to the data layers defined above, the following terms are used throughout our analysis and data presentation:

  • Metropolitan Statistical Area: One of 542 statistical regions in the United States generally defined as having more than 50,000 residents.
  • Micropolitan Statistical Area: One of 384 statistical regions in the United States generally defined as having more than 10,000 residents but less than 50,000 residents.
  • Number of Hospitals: Total number of healthcare institutions in a region that provide inpatient, therapeutic, or rehabilitation services, under the supervision of physicians.
  • Number of Hospital Beds: The sum of the hospital-reported "adult, pediatric, birthing room, or newborn ICU beds (excluding newborn bassinets) maintained in a patient care area for lodging patients in acute, long term, or domiciliary areas of a given hospital" in a region.
  • Number of ICU Beds: The sum of the beds in a region that fit the definition set forth in CMS, Section 2202.7, 22-8.2.

Team and Acknowledgements

Analysis and mapping work was completed by the Angstrom Bio Team:

  • Data Analysis and Map Construction: Jonathan (JP) Feldmann
  • Analytical Methods: Carlos Santos, Ph.D. and David States, M.D., Ph.D.
  • Map Project Conceptualization & Design: Forrest Cox

We are extremely grateful for the incredible, open work conducted by Johns Hopkins CSSE, the Harvard Institute for Global Health, Definitive Data, The New York Times, and others during this outbreak.

We are also thankful for the widespread culture of openness, collaboration, and of academic and analytical rigor that makes projects like this possible.

Angstrom Bio, Inc. applies novel nucleotide chemistry and machine intelligence to the study, diagnosis, and treatment of disease. Our core technology is a machine learning-driven platform for the computational optimization of antibodies and of antibody-like proteins. We seek to develop ever deeper models of understanding of the dynamics that govern protein-protein interactions, and to enable the true rational design of the right molecules for a given use case. For more information about us and what we do send us an email at info@angstrom.bio.

Change History

  • 2021-02: Changed the default field, as the 2-week comparison has become less useful. Due to the ongoing state of the pandemic and an explosion of map sources like this, often containing further testing and vaccinating information, we are no longer updating this map on a daily basis.
  • 2020-06-09: Added support for displaying new cases per ICU bed, as a measure of local risk
  • 2020-05-19: Added support for displaying detail information comparing regions in tables below the map.
  • 2020-05-11: Changed field controls on mobile devices for ease of use. Added historical new case data charts to tooltips.
  • 2020-05-07: Removed doubling rate; added change in new case rates. Adjusted display of tooltips and default fields for ease of use.
  • 2020-04-19: Added deaths per case metric
  • 2020-04-09: Added zoom support on the map
  • 2020-04-06: Added more historical data back to the first U.S. cases
  • 2020-04-04: Added new deaths and deaths per 10,000 people
  • 2020-04-03: Minor display changes and ability to link to specific layers of the map
  • 2020-04-02: Added Doubling Time (days) as a measurement of rate of growth
  • 2020-04-02: Re-processed map data to remove county/MSA areas that overlap the great lakes and coastal waters; added state outlines
  • 2020-04-01: Adjusted log color scale for population to start at 10,000, to better highlight variations
  • 2020-04-01: Updated datasets to include additional counties that are not mapped to geographic locations in source data
Cases:
()
Cases per 10,000 People:
New Cases per 10,000:
Change in New Cases (Past Week):
Cases per ICU Bed:
New Cases per ICU Bed:
Cases per Hospital Bed:
Deaths:
()
Deaths per 10,000:
Deaths / Case:
Population:
Daily New Cases