To learn about what Facebook Data for Good is doing in response to the COVID19 pandemic, click here.

COVID-19 Forecasts

Facebook AI builds adaptive models and collaborates with experts to help the world better understand the spread of the virus. The COVID-19 forecasts produced by the models gives researchers and public health experts information that can help them with resource planning and allocation and early outbreak detection. These forecasts are developed using public, non-Facebook data, and serve as a tool to support our global efforts to keep people informed as the pandemic evolves. Our data-driven methods achieve strong performance when compared to other state of the art forecasts. See the bottom of this page to download the data and to read more about the methodology.

Features

High Spatial Resolution

The forecasts are currently available for all counties in the United States. By sharing forecasts at a county-level, we are aggregating public data sufficiently to protect individual privacy while allowing public health officials to make locally relevant decisions.

Artificial Intelligence Expertise

We leverage a variety of AI techniques to produce fine-grained predictions that can capture rapid changes in a given area. Our adaptive models capture short-term trends and take correlations between regions into account. 

14-Day Timeframe

The forecasts are forward-looking - up to 14 days from each weekly update - and available for download via the Humanitarian Data Exchange.

Openness

In addition to sharing the projection data publicly, we’ve published a research paper that details the methodology and techniques we used to generate the forecasts.  

Who Uses COVID-19 Forecasts

You can learn more about Facebook’s COVID-19 forecasting efforts on the Facebook AI blog

New York University & Cornell University

Facebook AI has partnered with New York University’s Courant Institute of Mathematical Sciences to create localized forecasting models for the spread of COVID-19 (based on Multivariate Hawkes Processes and Neural Relational Autoregression). Our colleagues at NYU leverage this information in their models to estimate how progression of the disease will affect hospitals, bed and ICU capacity, in addition to local demand for ventilators, masks, and other PPE needs at a hospital and county level. Similarly, we have partnered with Cornell University using public data published by the State of New York to model the spread of coronavirus in New York.

Universitat Politècnica de Catalunya

We are partnering with the Universitat Politècnica de Catalunya to see how similar forecasting techniques can be applied in Europe. In the meantime, the team is using our U.S. forecasting data to better understand how the pandemic is evolving in different parts of the world. They provide periodic reports to the European Commission (EC) with analyses and predictions of the spread of COVID-19 in European and other countries, as well as the effectiveness of ongoing prevention efforts. Our U.S. county-level forecasts will now be integrated into specific EC-bound reports, which the research team will use to provide a more comprehensive understanding of global hotspots.

University of Vienna

In collaboration with the Faculty of Mathematics and the Data Science research platform at the University of Vienna, Facebook researchers are using AI to generate district-level projections of where and how quickly COVID-19 is spreading in Austria. We use public data shared by the Austrian government about confirmed COVID-19 cases to generate weekly seven-day forecasts. We provide these projections to our partners at the University of Vienna, who use this information to analyze trends and share results with health officials.

Methodology

To improve the quality and robustness of COVID-19 forecasts, Facebook AI developed a new neural autoregressive model that aims to disentangle regional from disease-inherent aspects within these datasets. Central to this model is its ability to account for relationships between different counties; for example, an uptick in one area can have an impact on predictions for adjacent or similar districts. This spatial approach allows Facebook researchers to train models toward solutions where knowledge about the spread of the disease in one area can improve predictions in a different area and thus borrow statistical strength across counties.

Learn more about the methodology and research paper here

Case Studies

Access to Data & Contact Us

The forecasts are available for download under permissive license via the Humanitarian Data Exchange by clicking below. For researchers interested in collaboration, please contact AIforecasting@fb.com.

Download this data

To personalize content, tailor and measure ads and provide a safer experience, we use cookies. By clicking "Accept All", you agree to our use of cookies on and off Facebook. Learn more, including about controls: Cookie Policy