Goldmark: Web tools help predict epidemics
We're entering a new era in health care, and it has nothing to do with the law the Supreme Court just upheld.
As everybody knows, we spend trillions of dollars treating people when they get sick. Prevention often is cheaper. But once you get beyond vaccination -- one of the most powerful tools for improving human welfare ever developed -- how do you do prevention?
Many of us have an annual checkup, but those are really more like state of the union assessments, aimed at picking up longer-term trends and emerging conditions like heightened blood pressure. An annual checkup won't do you much good if there's a sudden flu epidemic in your town -- or an outbreak of listeriosis, a bacterial infection which in 2011 claimed 30 lives in the Rocky Mountain states.
Over the past decade there have been efforts to gather information more systematically on the diseases for which people are admitted to hospitals, and the federal Centers for Disease Control compiles reports from labs, hospitals, clinics and physicians to track disease incidence. This has been valuable -- but it's a lagging indicator. There's a delay as the data is gathered, then compiled and analyzed. What you really want in the health field are leading indicators -- information that tells you what's happening before, not after, a problem becomes serious.
Now public health professionals think they may have found a leading indicator -- or more precisely, that the Internet may have presented them with one. It turns out that a lot of people who feel they might be sick enter their symptoms into a search engine to help them figure out if they're seriously ill. Many even guess at what disease they have.
Unwittingly, they're also performing a public service. A study at the Johns Hopkins Medical Center in Baltimore found that these search queries can serve as reliable leading indicators of disease incidence. There is a "really high correlation between these searches in the community and what we're seeing in hospitals," said Richard Rothman, co-author of the study.
An early sign in the listeriosis outbreak of 2008, whose origin was eventually traced to a cantaloupe farm in Colorado, occurred when a lot of people started searching fever, diarrhea and stomach discomfort. This allowed public health officials to get ahead of the curve and start warning others to beware of the danger and avoid certain foods -- all as the outbreak was just getting started rather than weeks after it had gained a serious foothold.
Another study, this one in the American Journal of Tropical Medicine and Hygiene, compared the sending of Twitter messages on symptoms with the formal reported case count of a cholera outbreak in Haiti that followed the 2010 earthquake there. The curve of the Twitter messages closely matches the number of confirmed cases from official sources -- but preceded the official count by seven to 10 days at the beginning of the outbreak. A week can make a big difference in containing the spread of such an epidemic and preparing people to take preventive measures.
The use of web-search information to predict disease is just one example of how data will transform the way we live in all kinds of ways. We're still early in the development of these new tools, and in the case of crowd-sourced symptom and disease self-reporting, the experts have clearly and sensibly warned against the potential for false trends.
But in the world of public disease control, early warning translates directly into health and lives, and the queries you enter electronically may wind up reducing the burden of disease -- and heading off an epidemic.
Google Flu Trends
This data, provided by Google Flu Trends, estimates flu activity using Google Searches. This graph compares activity over several years.