Behind the Model: Products

Nowcasting

CFA develops and uses nowcasting approaches to account for delays and incomplete reporting in recent epidemiological data. Nowcasts provide timely information to decision-makers and the general public by providing real-time estimates of disease surveillance metrics, which may guide resource allocation decisions and inform public health guidance. These metrics can include measures of disease burden and disease incidence, such as hospitalization rates, death rates, and case counts. Read more for an overview of how and why CFA uses nowcasting in data analysis.

Diagram illustrating a general nowcasting approach. On the day that the most recent data are received, the trend in ED visits appears to be decreasing. The next day, more reports come in and the data from the past days are updated. The following day, even more reports come in, and the trend now appears to be decreasing. Using snapshots of epidemiological datasets over time, we can measure how long it usually takes for reports to be recorded and then estimate a statistical distribution of this delay. Nowcasts produce estimates based on the current reported data and this distribution of historical delays in reporting.
Illustration of a general nowcasting approach. Initially when recent epidemiological data, in this example emergency department (ED) visits, are received, the data appear to be trending downward. However, in the coming days as more reports are received and the data are updated, an increasing trend is observed. This is due to delays in reporting, which nowcasting aims to correct for. Nowcasting approaches do this by estimating reporting delay distributions using snapshots of archived datasets over time, and then use this to correct recent estimates based on what is known about the distribution of reporting delays.
Keep Reading: Behind the Model: Nowcasting

Short-term Forecasts to Estimate the Hospital Burden of COVID-19

CFA oversees the COVID-19 Forecast Hub. The Hub solicits short-term forecasts of COVID-19 hospital admissions from teams across the United States and combines them into a single ensemble forecast. Read more for an overview of short-term forecast modeling and the benefits of ensembling forecasts from multiple teams.

Graphic of COVID-19 hospitalizations from November 2024 until December 2024, which forecasted models of hospitalizations over a two-week forecast horizon
Forecasts of COVID-19 hospitalizations from individual models and the COVID-19 hub ensemble model with a prediction interval for the COVID-19 hub ensemble model
Keep Reading: Behind the Model: Short-term Forecasts to Estimate the Hospital Burden of COVID-19

How scenario modeling can inform decision-making

CFA has developed scenario modeling capability that allows decision-makers to examine a range of possible outcomes under different assumptions about the future, providing insights for addressing specific policy questions and for general preparedness and planning. Read more for an overview of how scenario modeling helped inform CFA’s 2024-2025 Respiratory Season Outlook.

Line chart of historic COVID-19 hospitalizations with ribbons for two possible peaks based on different scenario models.
Two possible scenarios for peak COVID-19 hospitalization depending on timing of summer COVID-19 activity and input data source
Keep Reading: Behind the Model: How Scenario Modeling Can Inform Public Health Decision-Making

Supporting Decision-Making in a Local Measles Response

CFA is building modeling tools that can be used in real time to support decision-making for outbreak response. In March 2024, ÐÇ¿ÕÓéÀÖ¹ÙÍø’s National Center for Immunization and Respiratory Diseases (NCIRD) and CFA worked together to support the response by the Chicago and Illinois Departments of Public Health to a measles outbreak. The models helped forecast the expected size and duration of the outbreak and helped the Chicago Department of Public Health plan the public health response, allocate resources, and evaluate the impact of public health measures.

Figure 3. Schematic of a population represented by compartmental and agent-based models. This graphic illustrates two schematics based on a real-world population represented by figures on the lefthand side of the graphic. Colors represent disease states for both models: susceptible (grey), exposed (light orange), infectious (dark orange), and recovered (blue). People who already have immunity due to vaccination start the model as "recovered" (blue).
Schematic of a population represented by compartmental and agent-based models (ABM). Compartmental models group the population by disease state, whereas ABMs simulate how infection progresses at the individual level. Colors represent disease states for both models: susceptible (grey), exposed (light orange), infectious (dark orange), and recovered (blue). People who already have immunity due to vaccination start the model as “recovered.”   
Keep Reading: Behind the Model: How Disease Modeling Supported Decision-Making in a Local Measles Outbreak Response

Improving ÐÇ¿ÕÓéÀÖ¹ÙÍø’s Tools for Assessing Epidemic Trends

CFA is building modeling tools and computational pipelines so that we can analyze data quickly and accurately in response to outbreaks. Our goal is to make these tools accessible to federal, state, tribal, territorial, local, and academic partners. One of these efforts is to estimate the time-varying reproductive number, Rt, a measure that helps us quickly assess whether infections are increasing or decreasing.

Idealized Rt chart. Transmission chain of 3 infected generation persons to 4 infected generation persons.
Rt is a data-driven quantity. We identify all the individuals infected on a particular date (this is referred to as the "infectee generation", with four newly infected persons), and divide by the number of people who caused those infections (or the "parents" of those cases, making up the "infector generation").
Keep Reading: Behind the Model: ÐÇ¿ÕÓéÀÖ¹ÙÍø's Tools to Assess Epidemic Trends

Estimating Impact of Updated Isolation Guidance

We estimated the potential impact of ÐÇ¿ÕÓéÀÖ¹ÙÍø’s updated respiratory virus isolation guidance on an individual’s COVID-19 transmission potential. Our analysis found little difference on average for the likelihood that someone with COVID-19 would spread it to others under the updated isolation guidance compared with the previous guidance.

Blue mask outlined in black next to a coronavirus image with darker blue crowns.
Post-isolation precautions are a key feature of the updated guidance.
Keep Reading: Behind the Model: Estimating the Impact of Updated Isolation Guidance on COVID-19 Transmission

Wastewater-Informed Forecasting

CFA is building public health modeling tools that use signal fusion, in which data streams from multiple sources are used to produce more accurate modeling and analytics. In one example of this work, we are partnering with the National Wastewater Surveillance System (NWSS) and the National Center for Immunization and Respiratory Diseases (NCIRD) to use wastewater data alongside hospital admissions data to forecast COVID-19 hospital admissions at the state and national levels. These additional data could improve forecasts at critical times, such as when entering a surge in transmission.

Image of water flowing out of pipe
Monitoring pathogens in wastewater can help track community spread of COVID-19 and other diseases.
Keep Reading: Behind the Model: Wastewater-informed Forecasting of COVID-19 Hospital Admissions
Print
Content Source:
CFA, Behind the Model