What We Do

Center goals: Enable timely, effective decision-making to improve outbreak responses using data, modeling, and analytics.

Center for Forecasting and Outbreak Analytics Summary [PDF – 424 KB]

Early CFA Successes
  • CFA partnered with the pediatric acute hepatitis incident response team to produce a detailed technical report of the ongoing investigation. Learn more.
  • Although CFA was still in the pre-launch phase of development in winter 2021, the team pivoted quickly to anticipate the timing and impact of the Omicron variant on cases and hospitalizations in the United States. In partnership with Kaiser Permanente Southern California and UC Berkeley, CFA produced the first US estimates of Omicron severity compared to the Delta variant. CFA, in collaboration with teams in academia and experts in the office of the Assistant Secretary for Preparedness and Response, have also contributed analyses related to school test-to-stay polices, travel policies, and vaccine policy.
  • Within days of recognizing that the Omicron variant would cause a surge in the United States, the CFA team alerted federal leaders, state and local public health partners and the public that an impending increase in cases and hospitalizations would be severe enough to disrupt the functioning of critical infrastructure. This activity gave leaders several weeks of advanced notice of the timing and magnitude of the surge, allowing key planning activities.
  • CFA has already awarded $21 million in funding to academic institutions to advance modeling and forecasting methodology, with an emphasis on workforce development and health equity. Additional investments will be announced in the coming months.

The Center has 3 core functions:

Predict

Generate forecasts and analyses to support outbreak preparedness and response.

  • Model and forecast outbreaks and produce analyses.
  • Evaluate forecasts and other analytic products produced during an outbreak to assess performance and advance the state of the science.
  • Establish and maintain a data and analytics technology architecture.
  • Collaborate with federal, state and local leaders to support decision-making.
Inform

Share timely information with the federal government, state and local leaders, and the public.

  • Translate and communicate forecasts and analyses.
  • Maintain a network to engage decision makers, including public sector, private sector, and civil society.
Innovate

Advance research and development priorities to improve the performance of outbreak forecasts and analyses.

  • Identify, assess, and enhance foundational data sources and studies in collaboration with public, private, and academic partners.
  • Support the development of accurate, timely, and useful forecasts and analyses.
  • Establish a network of test beds among state and local jurisdictions to design, build, and test novel analytics and data sources.
  • Advance communication and visualization capabilities.
CFA's 101 for Industry Event

The Center for Forecasting & Outbreak Analytics (CFA) hosted this event to connect with industry leaders and share information about the new center. The agenda for the day includes a keynote speech from Dr. Nirav Shah, presentations by CFA leadership, and panels of industry leaders focused on technology relevant to disease forecasting, modeling, and outbreak analytics.  Event Video Part One | Event Video Part Two | Slides from Event [4.2 MB, 65 pages]

Engaging with State, Tribal, Local, and Territorial (STLT) Partners

CFA conducted several focus groups to engage STLT partners in providing feedback on CFA’s early analyses and technical products produced during the COVID-19 Omicron wave.

General Comments/Observations

  • CFA’s Role
    • Partners hope CFA will add to existing capacity and not aim to replace any existing communities or modeling staff in programs at CDC.
    • Partners expressed that it’s important CFA prioritize openness and transparency with STLT partners and the public.
    • Partners expressed interest in making the results of CFA’s forecasting and analytics more broadly available, especially to their jurisdictions.
    • Partners felt strongly about not wanting CFA to prioritize publishing, data hoarding, etc. during peacetime and/or times of crisis. Instead, focus on helping the public.
    • Partners expressed that scenario-based modeling is more useful that prediction forecasting.​

      • Scenario models from CFA can be a more flexible tool to understand future scenarios​

    • Partners see value in CFA developing and standardizing a tool to survey state or local populations to get clarity on public health issues (i.e., vaccine hesitancy) ​

    • Partners see value in a tool which can help predict hospitalization rates and inform surge planning needs​

  • Effectiveness of Tools
    • Partners felt that tools (such as nowcasting, reproductive number, descriptive analyses, etc.) that can conduct analyses at state and local levels will be a significant value-add, especially for public health departments who lack the funds/resources to develop them on their own.
    • Partners expressed that local governments appreciate having dashboards and that there is ‘significant’ use of dashboards at the local level to drive decision making.
    • Partners emphasized that the effectiveness of these provided tools are dependent on data collection and expertise within state/local health departments so it will be important for CFA to consider adaptability and regionality at state/local levels.
    • Partners felt that it’s helpful when CDC partners with stakeholders to conduct studies that can be extrapolated, especially for complex topics such as masking, vaccination, etc.
  • Barriers
    • Partners felt that one of the largest barriers to quickly reacting to outbreak events is the lag time between data collection and reporting, for example discrepancies in the data of specimen collection vs date of report.
    • Some participants were concerned with the granularity of data; more granular data is helpful at state/local levels, but not always available.
    • Participants were concerned with the lack of expertise/resources in smaller, rural counties. They recommended that ‘Plug and play’ model for data forecasting could address this issue (e.g., developing tools which don’t require knowledge of programming languages).

Suggestions/Recommendations

  • Integration
    • Partners expressed interest in having data tools developed in multiple “languages” to avoid issue of state/local health departments using different programming languages.
    • Partners emphasized a need for defining standards across state and local health departments (e.g., definitions, cut-offs, parameters) to ensure appropriate comparison.
    • Partners wished for an explanation of data inputs and assumptions of the model before disseminating.
    • Resounding belief that creating trainings and/or videos on interpreting a model will aid partners in providing effective, reliable resources for the public to utilize.
  • Data and Reports
    • Partners strongly value information on how interventions can be expected to impact case growth rate in some models.
    • Partners would like CFA’s commentary on the impact of mitigation efforts and potential policy implications with provided models.
    • Partners suggested adding other data sources such as tele-tracking, Google searches, and social media.
    • Partners shared a keen interest in implementing wastewater surveillance data to models
    • Partners believe shifting focus to hospitalization and deaths is more beneficial for forecasting as home tests become more common.
    • Partners would like CFA to think through how the center can develop upstream solutions to make data quality more robust and incorporate health equity factors​

  • Communication
    • Partners strongly believe scientific papers aren’t always the most streamlined form of communication. Rather, it is more helpful when CDC develops one-pagers and basic descriptions to be disseminated by public health departments.
    • Partners expressed an interest in holding regular science briefings with Q&A where pre-published studies can be highlighted and provide a more targeted focus on outbreak events.
    • Partners need communication support for politicized and/or controversial interventions and outcomes. They would value well-developed communications tools that show effectiveness of interventions/policy would be valuable.
    • Partners expressed interest in having liaisons from CFA to assist in tailoring communications to different audiences (e.g., data modelers and communication specialists)​

Page last reviewed: June 17, 2022