Most folks who wake up feeling crummy will sit down with a computer or smartphone before they sit down with a doctor.
They might search the Web for remedies or tweet about their symptoms. And that's why scientists who track disease are turning to the Internet for early warning signs of epidemics.
"Surveillance is one of the cornerstones of public health," says Philip Polgreen, an epidemiologist at the University of Iowa. "It all depends on having not only accurate data, but timely data."
Public health officials have been trying to speed up their responses to disease outbreaks since, well, they started responding to outbreaks.
There's still plenty of room for improvement.
The current system requires the Centers for Disease Control and Prevention to compile reports about from physicians and labs all over the country — and that can take a while. There's typically a week-long delay between an outbreak and the release of an official report.
To get an early read on things, epidemiologists look for the first clues of illness — a rise in thermometer sales or increased chatter on hospital phone lines. Now, they're tapping into the Internet.
ah . . . ah . . . AH-CHOOGLE
In 2008, Polgreen found a strong correlation between flu-related searches on and actual flu cases in the U.S. Around the same time, Google unveiled Google Flu Trends, which estimates flu activity based on Web searches.
Google Flu Trends
Google's estimates for the percentage of patients with flu-like illnesses closely match reports from the CDC.
The principle is pretty simple. On the Internet, people who type "fever" or "flu medication" into a search bar are more likely to end up in a hospital a few days later. Data collected from millions of these searches can be shared as they happen, many days before official reports.
But this speed comes with questions of accuracy. Can these signals be trusted? "Skepticism is healthy," Polgreen says. "Because there's a novelty effect, there may be excessive expectations both among researchers and the public."
With this in mind, a team of researchers at Johns Hopkins wanted to see if Google's estimates would prove accurate and useful in the everyday operation of a hospital. They compared Google searches originating in Baltimore to the number of patients who showed up with flu-like symptoms at a local emergency room.
"It seems like a stretch, but what we found — amazingly — is that there's a really high correlation between these searches in the community and what we're seeing in hospitals," says Richard Rothman, the study's co-author.
Search data were an accurate predictor of ER crowding. But that wasn't the study's most surprising tidbit. One weakness of online surveillance is that public hysteria about an epidemic might skew the results.
For example, when the bird flu outbreak was announced in 2009, there was a huge spike in flu-related searches in Baltimore. Public health officials labeled the phenomenon "fear week."
"Even though this was technically a false alarm, it did mirror what we saw in the emergency department," says Andrea Dugas, the lead author of the study.
There was a corresponding spike in ER visits. In fact, 6 percent more people showed up during the week of heightened anxiety about the flu than the week when swine flu reached Baltimore months later.
Search data aren't the only public health clues. John Brownstein and Rumi Chunara, researchers from Harvard Medical School, traced the cholera outbreak in Haiti by gathering data from all over the Web.
Instead of counting search terms, they collected thousands and thousands of Twitter posts, and used HealthMap, an online disease surveillance tool, to compile cholera-related news reports. The first signs of a cholera epidemic could be seen in these data two weeks before official reports. If officials had been monitoring the Web, they might have been able to respond faster.
"Still, there are drawbacks to these Web-based data streams," Brownstein says. The correlation between online trends and real life can falter when public interest wanes, he says.
Tracking Haiti's Epidemic Online
Official reports on Haiti's cholera outbreak closely tracked data from Twitter and HealthMap (an online surveillance tool created by John Brownstein, an epidemiologist, and Clark Freifeld, a software developer) for the first 100 days of the epidemic.
In the early days of the cholera outbreak, Twitter and HealthMap data closely matched official reports. After 100 days online activity dropped off even though the epidemic raged on. Disease is easiest to detecte when it is out of the ordinary. Regularly recurring maladies like malaria are hard to follow online.
Communicating About Disease
Online disease surveillance — or "Webidemiology" — is a cool new tool, and researchers are eagerly testing it out and double-checking the data they collect. But it won't be used by itself to make important public health decisions anytime soon.
"From an early detection standpoint there is a lot of work that needs to be done to sift through the noise," Brownstein says.
While researchers are working out the kinks of flu surveillance and a few other illnesses (Google just expanded its operation to track Dengue fever) other diseases will always fly under the Internet radar. Some have nonspecific or overlapping symptoms so they're hard to recognize. And there's a whole host of illnesses, such as sexually transmitted diseases, that people aren't necessarily eager to discuss on the Web.
"The Internet is just one additional stream of information," Polgreen says. "It's certainly not going to replace traditional forms of surveillance."
While the Internet may not be a perfect predictive tool, researchers and public health officials agree that it is great for one thing: communication.
Social networking allows officials to easily reach the public and enter into a conversation. Tweets, searches and Facebook posts can give officials a sense of public reaction to vaccines, or their attitude towards an epidemic.
"It's a quick and easy barometer for public anxiety," Polgreen says.
And in a public health emergency, that can be just as useful as cold, hard numbers about cases.