Zombies. They're everywhere. My dentist and his assistant spent my last visit and chatting about The Walking Dead while drilling into my head, and it seems like every reasonably large town hosts a zombie run. Science education is getting in on the trend, too. Colleges have classes about zombies, AMC (the network that broadcasts The Walking Dead) is sponsoring a free zombie-themed survival course, and Texas Instruments developed an app for teaching neuroscience using ideas from zombie movies. Even the CDC wants to help you prepare for the zombie apocalypse. Frankly, it's a little out of hand. How long will this zombie fever last?

Sarah Reehl, who is now a mathematics graduate student at Utah State University, noticed the ubiquity of zombies while she was working on her undergraduate thesis at Carroll College in Montana last year. "I was watching The Walking Dead and trying to make physiologically correct arguments about zombie mechanics. My instincts were to model a real zombie infection," she wrote in an email. But in her search for data about zombie behavior, she decided that zombies were more interesting as a pop culture phenomenon and shifted her focus. In her thesis (pdf), Reehl used a standard epidemiological model to study not only seasonal influenza rates but also our culture's obsession with zombies.

Reehl used what is called an SIR model for both influenza and zombie fever. SIR stands for three groups in a population: "susceptible," "infected," and "recovered" or "removed." It is particularly useful for diseases—such as measles, chicken pox, and influenza—that people typically get only once. (Yes, you may have gotten the flu more than once, but you didn't get the same strain every time. Influenza mutates so quickly that long-term immunity is impossible.) In disease modeling, the population consists of people or other organisms, and researchers determine a system of differential equations that represent the likelihood of transmission of the disease and how quickly an epidemic will pass through the population.

"What happens if we try to model popular culture trends as infectious diseases?" Reehl asks in her thesis. With the help of Google's n-gram viewer, she used words instead of people as the population in question. The n-gram viewer is a tool for exploring trends in language. When you search a word (a 1-gram) or phrase (an n-gram, where n is the number of words in the phrase), you get a graph that shows its popularity over time, based on its prevalence in a corpus of books that Google has digitized.

Reehl looked at two past linguistic phenomena, "jitney" and "groovy," to get a feel for the rise and fall of trendy words. Based on this background, she developed an initial guess of a model for "zombie" prevalence, which she then refined. Of course, the devil is in the details, and different assumptions about transmission rates and the susceptibility of the language to the word "zombie" yield different models. Reehl believes that in the most likely model of zombie fever, over the next 25 years, "zombie" will come close to the popularity of Halloween stalwarts "vampire" and "Frankenstein," after which it will gradually subside but not die out completely. But she also thinks it would be interesting to develop a model that would take into account the use of the word "zombie" not only in books but also on Twitter and television, where she thinks zombies might be more popular.

If you, like me, are too squeamish to watch zombie movies and are therefore kind of bored with the trend, it's going to be a while before the last axe settles into the skull of a fictional monster with a hunger for brains. Reehl predicts that zombie fever will peak around the year 2038, but it will probably not return to pre-Shaun of the Dead levels until after 2080. Better keep the axe handy.