Outbreak Response Modeling

What to know

  • CFA supports CDC outbreak response by using modeling to answer a wide range of questions that are important to public health decision makers and outbreak responders.
orange background with blue waves and white page title text

Modeling can help answer important questions

Modeling can be used in infectious disease outbreaks to help answer a variety of questions that are important to public health decision makers. Models leverage available data and our understanding of infectious diseases to provide insight into both the present and the future of an outbreak.

What is happening now?

There are many reasons that it may be hard to know the size of an outbreak while it is happening. At the beginning of an outbreak, there may not yet be a clear case definition, disease surveillance program in place, or data reporting processes. Even after these elements of outbreak response have been established, reporting delays are often unavoidable, which can make it look like outbreaks are slowing even when they are still getting worse. CFA uses a variety of modeling approaches to address these types of problems:

  • CFA used a branching process model early on in the 2025 Marburg outbreak in Ethiopia to provide estimates of how many total cases there may already be, based on how many cases had been identified and reported. This was shared with CDC responders to help understand the possible scope of the outbreak.
  • During the 2025-2026 measles outbreak in South Carolina, CFA used nowcasting to adjust for reporting delays, providing more accurate case counts to local public health officials.

What is coming next?

Decision makers often want to use modeling to look into the future, to know if an outbreak will happen, how big it will get once it's underway, and how long it will last. Questions about the short- and medium-term future of an outbreak can be answered using a variety of methods. While we have methods for short-term forecasting of seasonal respiratory diseases, these models usually rely on longer-term, higher-quality data than what are typically available in a new infectious disease outbreak. Instead, we often use other types of modeling, such as transmission models, and fit them to what has happened so far in the outbreak to make predictions about where the outbreak is headed.

How should we respond?

Modeling can be used to help guide or support interventions in an outbreak. This often falls within the general domain of ‘scenario modeling,’ because models are used to compare different scenarios. For example, a situation in which a vaccination campaign was conducted could be compared to a hypothetical one in which it wasn’t. Modeling can be particularly useful for these types of questions because they compare different possible futures, or even a scenario that did happen with one that didn’t. These models can be used to help support decision making within the context of an outbreak.

  • As part of CDC's 2025 measles response, CFA developed the Measles outbreak simulator, a compartmental modeling tool that allows users to simulate outbreaks of measles in their own communities and look at how the outbreak changes under different levels and combinations of infection control interventions.
  • In response to the 2025 outbreak of Sudan Virus Disease in Uganda, CFA used a branching process model to illustrate the potential impact of contact tracing program effectiveness on the trajectory of the outbreak. This work was used to support the conclusions of an outbreak scenario assessment.
  • During the 2024 monkeypox response, CFA used a network model to demonstrate that transmission of Clade 1 monkeypox within households was unlikely to be of major concern, resulting in very small outbreaks in simulations, and that the disease was unlikely to spread between households. This work was shared with CDC decision-makers and published in a technical brief.

Modeling Approaches

Nowcasting and Rt Estimation

Reporting delays can pose challenges to understanding recent trends in an outbreak. Nowcasting addresses this by adjusting incomplete data based on historical reporting patterns, improving situational awareness and aiding decision-making. While nowcasting is not always possible in an outbreak, it can be used in outbreak situations where there is regular reporting of case data. Nowcasting can also be used to estimate Rt, the time-varying reproduction number, which indicates whether an outbreak is growing, declining or remaining steady. It is a useful measure for understanding epidemic trends and measuring the impact of interventions.

Branching Process Models

Branching process models are a type of stochastic, agent-based transmission model that shows how an outbreak may grow over generations of disease transmission. They are often used at the beginning of an outbreak. Compared to other modeling approaches, they often require fewer parameters, which makes them easier to implement with limited information. Additionally, they incorporate random variation, which is particularly important early in an outbreak.

Compartmental Models

Compartmental models represent populations as a collection of groups with similar transmission characteristics. These models can be constructed from just a few compartments where broad groups of individuals have the same characteristics, or from many compartments that capture unique characteristics of each group. They can be designed to incorporate different interventions that impact transmission rates, such as vaccination, quarantine, or isolation. We can use these models to understand the spread of disease both within and between populations.

Network Models

Network models are individual-based models that incorporate specific contact structures between individuals. CFA has used network modeling to answer questions about the types of contact that could lead to outbreaks. Network modeling is most useful when there's detailed information on contact structures in the population(s) of interest.

Print
Content Source:
Center for Forecasting and Outbreak Analytics