Five Facts about Epidemiological Models

Researchers rely on various epidemiological models to predict the way pathogens such as the coronavirus might spread in given areas. Different models can produce strikingly different numbers on how many people might fall ill and die, because they’re based on frequently changing facts and assumptions.

  1. Numerous models worldwide make different projections about the coronavirus’ spread.

Various nations, states, universities and institutes conduct models that produce a sometimes-confusing array of projections. Among the first to publish was a team of Hong Kong and China-based researchers, in January. Their widely read article said every infected person at the outbreak’s early stage could reasonably be expected to infect two to three others. The White House sometimes cites a model conducted by the Institute for Health Metrics and Evaluation at the University of Washington School of Medicine. Colorado is among the states producing their own, localized models.

  1. A London-based model alarmed and energized U.S. officials and others.

A mid-March epidemiological model from Imperial College London projected that without drastic interventions, Covid-19 would kill more than half a million Britons and perhaps more than 2 million Americans. The report shocked many at the White House and elsewhere, and helped spur new government actions such as calling for “social distancing” and widespread business closures.

  1. Epidemiological models rely on assumptions, but they’re not all the same.

In an epidemic’s early stages, when hard data is scarce, researchers rely heavily on mathematical models that predict where potentially infected people are going and how likely they are to spread the disease. Many of the models “are unique to individual academic groups that have been developing them for years,” the journal Nature reports. As time passes and data comes in, researchers constantly update their models, replacing earlier assumptions with new facts.

  1. A model’s projections may change sharply as facts and human behavior change.

The Imperial College report shocked Britons into dramatically changingtheir movements and other behavior. Neil Ferguson, the model’s chief researcher, soon downgraded his projection of U.K. deaths to a much lower number, 20,000. Some journalists questioned his reliability, but Ferguson explained that models are meant to react and update as facts change. He noted that the projection of more than 2 million U.S. deaths was based on the assumption that Americans would not change their behavior, which quickly became outdated starting in mid-March.

  1. Models are not gospel.

Researchers warn that epidemiological models are snapshots in time, and not meant as sure-fire, long-lasting predictions. “Don’t obsess about the numbers because they will change,” says Dr. Ross E.G. Upshur, head of clinical public health at the University of Toronto’s Dalla Lana Faculty of Public Health. They are important planning tools, Upshur says, but “all models lie” (a popular expression in his professional world). Some models, he adds, “are more useful than others.”


become a member       become a member       become a member       become a member       become a member      
Scroll to Top