Wikipedia is great for looking up who won the World Series or Super Bowl in a given year or details about the details depicted in a movie like American Hustle after seeing the flick. The site is generally not used as an official source by researchers and publications (Guardian Liberty Voice does not use Wiki as a source). However, Wikipedia use is turning into a great prediction tool for public health officials trying to gauge the spread of the flu, other viruses as well as growing health concerns.
People typically turn to the Internet for health information to self-diagnose an illness by checking on symptoms. In fact, studies show 72 percent of Americans sought health information online in the last year.
Now, public health experts are using online activity to map diseases as they spread. Healthmap.org and Google Flu Trends, for example, analyze online search activity to create almost real-time maps of diseases. (Note, some search rates based on media activity do not reflect infections, such as people looking for info on Ebola in the past few months.)
A study just published in the journal PLOS Computational Biology illustrates how activity in Wikipedia was used to track – and predict – flu outbreaks. The study and article show how researchers from the government’s Los Alamos National Laboratory facility used Wikipedia page views as a prediction tool.
The research team developed a list of Web pages that people jump to from the Wiki “influenza” information. They compared that Web traffic with reports issued by the U.S. Centers for Disease Control and Prevention (CDC) and the World Health Organization. They narrowed their original list of pages people go to on Wikipedia to 10 flu-related topics. They then used those pages to develop an algorithm to track usage.
Google or other sites that keep data and search information confidential. However, Wikipedia provides public access to its traffic data as part of their mission to make information openly available. Employed effectively, that data can be used to predict seasonal diseases.
The researchers did find that media activity affected the data and that studying diseases with long incubation periods or that are not seasonal in nature, such as HIV, did not really work. But they determined that their algorithms could spot health trends on seasonal viruses before they become large outbreaks in an area. By contrast, the CDC ‘s weekly flu report is published two weeks after the information is gotten on influenza rates from hospitals and doctors.
One shortcoming of the Wikipedia health data extraction is the difficulty in determining geographic information. The researchers admit that just figure that someone pulling up information in German is in Germany. That logic works for languages that are used regionally but is surprisingly imprecise for others. For example, information sought in Spanish could be a data search in parts of the U.S., Mexico, most of Central and South America, Spanish, etc. Such data would just show global instances, but not be a predictor of flu outbreaks in South America,
Even with the drawbacks, the researchers believe that Wikipedia can be used in the future as a public health gauge. If successful, a global disease-forecasting system could possibly change the way health officials respond to epidemics, according to lead study author, Dr. Sara Del Valle, from Los Alamos. She believes that eventually people (will) monitor disease prevalence trends and plan for the future based on the forecast, much like people check the weather forecast for the next week.
By Dyanne Weiss