You're sitting at lunch when your friend hands you a note with some bad news: You've been infected with Muizenberg Mathematical Fever (MMF). Are you going to get sick? Will you die? To find out, you visit a website that reveals the severity of your infection and how many people you'll infect. As the outbreak spreads among your colleagues, some report to the health clinic. Others go untreated. Fortunately for you, this is all a simulation. It's part of a new game designed to teach students the complexity of data generated by outbreaks.
MMF is the brainchild of Steve Bellan, an ecologist at the University of California, Berkeley, who specializes in the epidemiology of wildlife diseases like anthrax, and Juliet Pulliam of the University of Florida, Gainesville. The pair teaches at the annual Clinic on the Meaningful Modeling of Epidemiological Data (MMED) in South Africa, a 2-week program designed to provide mathematicians—mostly from around Africa—with broad lessons about epidemiology. The mathematicians, Bellan says, are topnotch, but often don't have a lot of training in study design and infectious disease data analysis. Mathematical models of infectious diseases play an important role in informing public health policy. But modelers should consider questions like "How are patients who report symptoms different from those who hide them?" or "Were some people more likely to be included in the study than others?" Ignoring these aspects of the data can lead to major biases and misleading models. The program aims to teach these concepts along with mathematical modeling and, most importantly, how they both fit into the big picture of epidemiological research, Bellan says. The clinic has been going on for 4 years, and last year he and Pulliam developed the MMF simulation as a teaching tool.
Here's how the simulation works. Most mathematicians in the workshop are "infected" by fellow attendees whom they interact with. Like a real disease, MMF spreads among the people who spend the most time together. When people receive their diagnosis, they go to a website that gives them more information. On the site, a random number generator determines whether they'll have symptoms and how many others they'll infect.
The students are also told to inform the instructors of their infection. But only some are also instructed to visit a makeshift health clinic. The result of the game is two data sets: The instructors have the omniscient data set that records every infected person, which would never exist in the case of a real disease outbreak. There's also a more realistic data set consisting of cases reported to the health clinic.
"When you have a real outbreak, you need to rely on whatever data you can get," Bellan says. "And our exercise helps show the difference between that data and the reality of the outbreak."
When MMF has run its course through the workshop, the mathematicians work with the data set from the health clinic to make predictions about how MMF spreads, what influences its severity, and why the outbreak burns out. They can see how far off their predictions are from reality by checking against the omniscient data set. The instructors can create different versions of MMF with different modes of transmission—environmental versus person-to-person, for example—or different rates of infectiousness. When the workshop moves into its second week, consisting of group projects on data analysis of any diseases, many groups choose to continue working with MMF. The students say that understanding how the data set was generated helps them mathematically model the data more effectively, says Bellan, who describes the details of the game today in PLoS Biology.
"The feedback we've gotten from those who have gone through this has been very positive," Bellan says. And instructors from the 2-week workshop are now taking it back to their home institutions, testing it out with different types of students, he says: "We think in large undergraduate classes, this would be very successful."
Gavin Hitchcock, a specialist in epidemiological pedagogy, or teaching, at the South African Centre for Epidemiological Modelling and Analysis in Stellenbosch, South Africa, who was not involved in developing the MMF program, says that bringing together different methods of data analysis is a challenge in the classroom. The new simulation, he says, "brilliantly exemplifies how this integration may be achieved in a practical, hands-on, memorable way." It could be useful for students from high school through graduate school, he adds.
Bellan and other founders of the program plan to continue creating different versions of MMF and expanding it to cover more aspects of epidemiology. Some areas of investigation include figuring out what happens when two vastly different strains of a disease present with similar symptoms, and what to do when a disease is spreading by both person-to-person contact and a water or food-borne pathogen at the same time.
Correction: This article has been amended to reflect the following: Juliet Pulliam and Steve Bellan contributed equally to the creation of the MMV simulation. In the last paragraph, the word "immunology" has been replaced with "epidemiology." Lastly, the second paragraph has been modified to reflect the fact that mathematicians who attend the 2-week MMED clinic are not doing incorrect work before they attend the clinic, nor do they have a gap in their knowledge that prevents their models from being accurate prior to the clinic; rather, they want to strengthen their models and their collaborations with epidemiologists.