Slide 1Mathematical models use equations to represent how diseases spread through populations, and they're a go-to approach when an experiment can't answer the question.

Slide 2When the Zika virus started spreading through Latin America in 2015 and 2016, what did public health officials need to know most urgently? Not just where cases existed that day, but where the outbreak was likely to spread next week and next month. Which countries should prepare for surges in cases? How many diagnostics would they need? We can't run an experiment to answer those questions, and by the time we've observed the natural course of events, it's too late to prepare.

Slide 3Think of models as simplified versions of reality that capture the essential dynamics while ignoring unnecessary details. The most common framework is the SEIR model, which divides a population into compartments: susceptible, people who can catch the disease; exposed, infected but not yet infectious; infectious, able to transmit; and recovered, immune. People flow from one compartment to the next based on parameters like the transmission rate, incubation period, and recovery rate. Change those parameters, and we change how the epidemic unfolds.

Slide 4When Zika hit, researchers built dynamic neural network models that could predict geographic spread in real time. These weren't simple SEIR models. They incorporated weekly case counts from Pan American Health Organization data, passenger air travel volumes between countries, mosquito habitat suitability, and socioeconomic indicators like GDP, population density, and healthcare infrastructure.

Slide 5The model took all these data streams, normalized them, and used them to forecast which countries would become high-risk in the coming weeks. Even when predicting twelve weeks ahead, the model achieved over 85% accuracy. That meant officials could plan resource allocation not just for next week, but for the next three months.

Slide 6Not all models are created equal. In deterministic models, everyone follows the equations exactly, so running the model twice with the same starting conditions gives identical results. That's good for understanding general trends. Stochastic models build in random variation, reflecting the reality that epidemics don't unfold in perfectly predictable ways, and they're better for capturing uncertainty. Agent-based models track individuals instead of populations. Each agent, a person or a mosquito, has its own characteristics and follows behavioral rules, and when thousands of agents interact, population-level patterns emerge.

Slide 7During West Africa's Ebola epidemic, modelers generated weekly forecasts for specific regions, publishing them online as new data arrived, and these forecasts informed real resource allocation decisions. The short-term forecasts, one to two weeks ahead, proved quite reliable, but forecasts beyond three weeks became increasingly unreliable.

Slide 8Why? Because human behavior changes in response to epidemics. People start avoiding handshakes, funerals change, and clinics adapt protocols. The models that worked best incorporated these behavioral changes through dampening mechanisms that constrained how fast case numbers could grow.

Slide 9What does a dampening mechanism actually mean in practice? A simple SEIR model might predict exponential growth, where each infected person infects others, who each infect more. But as an epidemic grows, people respond. They change their behavior, clinics improve infection control, and communities modify burial practices, and all of these responses slow transmission. Models that incorporated decay terms, or contact patterns recognizing that not everyone mixes with everyone else equally, reduced forecast error by approximately 90 percent.

Slide 10The practical implication is to use models for near-term operational planning, and not to expect them to predict the distant future with precision. Update forecasts constantly as new data arrive, and calibrate the model repeatedly as we learn more about how the epidemic is actually unfolding. The Ebola forecasters didn't just assess whether their predictions were close to reality. They used probabilistic calibration assessment to evaluate how well they had quantified uncertainty, because getting the uncertainty bounds right matters as well as the point estimate.

Slide 11Models can attempt to answer what-if questions that are hard or impossible to address experimentally, integrate data from multiple sources, and reveal dynamics like tipping points and feedback loops that aren't obvious from empirical data alone. But models are only as good as the assumptions built into them. Their outputs are projections under specific assumptions, not predictions of the future, and uncertainty increases with the forecasting horizon. A model that hasn't been validated against out-of-sample data may be fitting noise, and models that treat human behavior as static will systematically overestimate outbreak size. Always run sensitivity analyses to identify which parameters drive the results and where better data would improve the projections.