When the next epidemic comes, there’s a good chance it will switch flights at John F. Kennedy Airport in New York.
Researchers at MIT have developed a pretty nifty computer model to figure out the most influential airports in the early stages of an epidemic’s spread.
JFK is No. 1, followed by Los Angeles International Airport. You might be surprised to learn that Honolulu International Airport ranks third.
Why Honolulu? It’s a pipsqueak compared to JFK when it comes to traffic. But, Honolulu handles a lot of flights to and from Asia. And the people who move through the Hawaiian airport tend to travel a long way, taking germs with them. The flights in and out of Honolulu also connect with lots of major hubs.
Traffic does count in the computerized model. But traffic alone doesn’t predict the influence of an airport on a disease’s spread, the researchers say. The MIT model also takes into account layover times, the geographic location of airports and how networked they are with other airports.
So, according to MIT’s model, Hartsfield-Jackson Atlanta International Airport, the busiest airport in the world, comes in eight on the list of early, influential spreaders.
A relative few of the 40 airports analyzed were found to be “super-spreaders,” according to study published in the online journal PLoS ONE. Besides JFK, LAX and Honolulu, the other key U.S. airports are San Francisco International, Newark Liberty, Chicago O’Hare and Washington Dulles. (See the chart below, paying closest attention to the triangles toward the top.)
So what does it all mean?
“Our findings could perhaps point in two directions,” MIT geoscientist Ruben Juanes told Shots in an email. “One is that it identifies some surprising pathways for aggressive spatial spreading (like Honolulu), and this type of information could be used to allocate intervention strategies (like vaccinations) ahead of time.”
And where else? “The other avenue that our study points to is that by refining this type of model, one can make real-time forecasts of geographic spreading for specific outbreaks based on realistic air traffic, travel patterns, and individual-mobility behavior,” he wrote.
Juanes, whose lab spends more time figuring out how fluids flow through soil, says the model is pretty generic and could apply to quite a few contagions, including flu.