What if we were able to know exactly when, say, the next bird flu pandemic was going to hit—that two weeks before it appeared, scientists would have seen it coming and been able to prepare a vaccine and start the production and distribution processes to get the drug to the people who might need it. Great, right?
Worldwide, scientists are trying to do just that—and they’re getting closer to being able to do it all the time. In the U.S., the Centers for Disease Control and Prevention (CDC) publish the Morbidity and Mortality Weekly Report (MMWR), which is the agency’s primary verifiable source for news on infectious disease outbreaks. Unfortunately, the information in it is only made available two to three weeks after an outbreak occurs—during which time an outbreak can grow into an epidemic, or peter out completely. While this formal data ensures the accuracy of information, it doesn’t provide real-time information on the national or global state of infectious diseases.
At HealthMap, our focus is on early detection of emerging infectious diseases. Through the use of informal sources we aim to discover outbreaks as soon as they occur, enabling authorities and medical professionals to quickly respond and contain the outbreak. But imagine pushing these methods one step further: actually predicting a disease outbreak before it happens, using existing data and a little bit of educated guesswork. Many researchers are doing exactly that, forecasting when and where a disease will emerge, as well as the magnitude of an outbreak. Here are three ways we look into our version of a Magic 8-Ball to see where a pandemic is likely to strike next:
Social media platforms like Twitter give real-time data that can identify disease outbreaks and also predict the timing and location of new outbreaks. Sure, there’s a lot of useless “noise” in the Twittersphere—outbreaks of Bieber Fever, for example, or mysterious zombie viruses (don’t ask). But when filtered correctly, the Twitter stream is a valuable resource for tracking diseases because tweets allow some description and provide context for time and location.
Similarly, Twitter profiles may contain important demographic information, such as location, gender, and age. Numerous studies have harnessed Twitter’s expansive network for real-time disease surveillance and this research shows that the information is timelier and as accurate as what’s collected by formal outlets like the CDC. Mark Dredze, an assistant research professor of computer science at Johns Hopkins University, in Baltimore, developed an algorithm that can automatically filter tweets to identify people sick with the flu and separate those from tweets that include the word “flu” but aren’t related to actual flu cases. When applied to the flu epidemic of January 2013, Dr. Dredze and his team matched the CDC’s estimates of actual flu rates. In 2009, researchers from Southeastern Louisiana University were able to accurately forecast future influenza rates by tracking rates of influenza-related messages on Twitter.
AIR TRAVEL PATTERNS
While word of infectious disease may go viral online, actual pathogens still generally require some sort of human contact to spread. Dirk Brockmann, an associate professor of engineering sciences and applied mathematics at Northwestern University, in Chicago, studies modern human migration and travel patterns to predict the start and spread of epidemics and pandemics. Using the air transportation network as his foundation, Dr. Brockmann has identified which airports and connections were the most heavily trafficked. With this information, he created a predictive computational model that redefines the notion of distance.
In the 21st century, with the introduction of increasingly complex transportation networks, diseases no longer spread according to geographic distance. Places like New York City and London are geographically far apart, yet the amount of air traffic between these two travel hubs makes them effectively neighbors. “The amount of traffic is an indicator or measure of how close different places are,” explained Dr, Brockmann in an interview with The Disease Daily. This traffic-based notion of distance can predict the arrival time and location of an outbreak.
Weather is, of course, a much older way to predict disease outbreaks.. You’ve probably noticed that you seem to get a cold or flu more often when it’s cold outside. A group of researchers at the National Institutes of Health’s Fogarty International Center recently published a paper examining the consistency of the relationship between seasonal flu activity and climate. It turns out that in temperate regions, cold and dry weather is associated with annual peaks in influenza cases. But tropical and subtropical regions (think Brazil, Senegal, or Indonesia) are more likely to experience a peak in flu cases during the rainy season. It’s important to remember that an association does not mean that cold or rainy weather causes the flu.
There are many different hypotheses for why weather and the flu are linked, such as decreased sun exposure inhibiting our immune systems; increased contact with other people when we’re stuck inside; and even the academic calendar (school-age children are considered “super-spreaders” of illness, for good reason). Whatever is behind the link, study authors believe these models could help predict when peak influenza activity will occur and allow public health professionals to better prepare.
Many vector-borne diseases—those transmitted by blood-sucking insects or arachnids, such as mosquitoes, fleas, or ticks—are closely associated with weather patterns. Dengue is a virus transmitted by the Aedes aegypti mosquito. Rainfall and temperature affect the lifecycle of A. aegypti, which in turn impacts disease transmission. Anna Buczak and her colleagues at the Applied Physics Laboratory at Johns Hopkins University examined the relationship between these factors to form a set of rules that predict whether the number of new cases of dengue in future weeks in Peru and the Philippines would be high or low. It turned out they could pretty accurately predict the number of new cases of dengue, theoretically allowing for earlier interventions and decreased impact of the outbreak.
Disease detection, particularly what’s called "digital disease detection," is a rapidly growing field. Several research groups are using novel methods of detection; the methods we’ve described here are just a sample. It’s clear that there’s no single best way to predict where and when the next big disease outbreak will occur. The emergence of a disease is complex and each has its own set of drivers, depending on the pathogen, its mode of transmission, vector life cycle, and more. No single strategy is able to cover all the factors that drive pandemics, but as methods develop, we get closer to finding and preventing the next big thing in infectious disease.
Do you fear outbreaks of flu and other viruses? Do you take any precautions when you travel or during certain times of the year when outbreaks may be more likely?
Related Stories on TakePart:
The Disease Daily is created by a team of medical doctors, veterinarians, and public health professionals who believe that infectious disease news should be accessible and comprehensible to everyone. As a publication from HealthMap at Boston Children's Hospital, The Disease Daily has access to real-time reporting of infectious disease events all over the world. While HealthMap alerts thecommunity to the outbreaks, The Disease Daily puts those alerts into context, showing readers the impact of infectious disease on policy, economics, and community.