Study: Wikipedia Views Can Accurately Predict Disease Outbreaks
November 15, 2014 in News
Wikipedia can be used to accurately predict the spread of certain diseases, such as influenza and dengue fever, according to a study published in PLOS Computational Biology, BBC News reports (BBC News, 11/13).
Details of Study
Researchers from Los Alamos National Laboratory analyzed Wikipedia traffic from March 7, 2010, through Feb. 1 to see how the site’s page views correlated with the spread of diseases.
Specifically, researchers compared a country’s governmental outbreak reports for a specific disease with the number of Wikipedia searches in the country’s language for that disease (Generous, PLOS Computational Biology, 11/13).
They examined a total of seven diseases across 11 countries (Izadi, “To Your Health,” Washington Post, 11/13).
Wikipedia visits proved to be an accurate indicator of outbreaks in eight of the 14 disease-location pairings, BuzzFeed reports (Phillips, BuzzFeed, 11/13). In addition, researchers found that the model could use data from an outbreak in one country to predict the same disease’s spread elsewhere.
Specifically, researchers were able to correctly predict:
- Influenza’s spread at least 28 days in advance in the U.S., Thailand, Poland and Japan;
- Dengue’s spread at least 28 days in advance in Thailand and Brazil (“To Your Health,” Washington Post, 11/13); and
- Tuberculosis cases in Thailand (BBC News, 11/13).
Government data generally lagged one to two weeks behind diseases’ spread and about four weeks behind Wikipedia search trends, according to the Washington Post‘s “To Your Health.”
Sara Del Valle, a project leader at Los Alamos who worked on the study, said the model was successful to the fact that people likely read about diseases online before visiting a doctor (“To Your Health,” Washington Post, 11/13).
Wikipedia Views Didn’t Indicate All Outbreaks
However, while the model was a successful predictor in some areas, the researchers found it did not forecast the spread of:
- Cholera in Haiti;
- Ebola in Uganda and the Democratic Republic of Congo; and
- HIV/AIDS in China and Japan.
The researchers attributed this to:
- Limitations of Wikipedia data;
- Diseases changing too slowly;
- Low incidence of the disease in the location studied; and
- Poor Internet connectivity (Buzzfeed, 11/13).
Additionally, the researchers noted that the study could be limited because certain languages are spoken in multiple countries (“To Your Health,” Washington Post, 11/13).