COVIDcast is the name for Delphi’s project in support of the COVID-19 response in the United States. It has several parts.
We procure unique data streams that reflect COVID activity from a wide variety of sources, and extract COVID-related signals. These signals cover every nearly every rung of the severity pyramid, and constitute what we believe may be the nation’s largest public repository of real-time, geographically-detailed indicators of COVID activity in the U.S.
We make all COVID-19 indicators freely available through a public API, which is updated with new data daily. We also provide associated R and Python packages, and a dashboard to visualize a core set of indicators.
Some of the underlying data sources (on which these indicators are built) would not exist or be publicly available without Delphi’s efforts. These include a massive national daily survey called the COVID-19 Trends and Impact Survey (CTIS) that we’re running in partnership with Facebook, which has been taken by over 25 million Americans since April 2020; and an enormous database of de-identified medical insurance claims, covering more than 35% of the U.S. population, made possible through health system partners including Change Healthcare.
We support and advise the U.S. CDC (in collaboration with the Reich Lab) on short-term COVID forecasting, which includes creating and evaluating an ensemble forecast out of the models submitted to the COVID-19 Forecast Hub, serving as the basis for the CDC’s official forecast communications.
A summary of some key parts of COVIDcast can be found in the December 2021 Proceedings of the National Academy of Sciences Special Feature: Beyond cases and deaths: The benefits of auxiliary data streams in tracking the COVID-19 pandemic:
More recent work on nowcasting using COVID-19 indicators: