A researcher has used social media to track attitudes about vaccination and how they correlate with vaccination rates, in the process creating a novel model to track a variety of disease states.
The study adds to a growing body of evidence that social networking can be used to track diseases and other natural disasters that affect public health. Earlier this year, researchers used Twitter to track rapidly-evolving public sentiment about H1N1 influenza, and found that tweets correlated with actual disease activity. Before that, researchers analyzed how Twitter was used to disseminate information (and misinformation) about flu trends.
In the latest study, published at PLoS Computational Biology, a biology professor at Penn State University compiled 477,768 tweets with vaccination-related phrases from August 2009, when news of a new H1N1 vaccine first was made public, and continued through January 2010.
According to a press release, students rated a sample of about 10% of tweets as positive, negative, neutral or irrelevant. The human-rated tweets were used to program a computer algorithm to analyze the remaining 90% of tweets, leaving 318,379 tweets expressing either positive, negative or neutral sentiments about the H1N1 vaccine. (Tweets found to have nothing to do with H1N1, such as messages about another vaccine entirely, were discarded.)
Of the remaining messages, 255,828 were classified as neutral, 26,667 as negative, and 35,884 as positive. Starting from late August 2009, there was a steady increase in the number of relevant tweets until early November 2009, after which the number dropped back to previous levels.
Negative expressions spiked during the time period when the vaccine was first announced. Later, more-positive sentiments emerged when the vaccine was first shipped across the United States. The researcher also tracked spikes of negative tweets that corresponded to periods of vaccine recall.
Also, the researcher modeled how social media users with either negative or positive sentiments about the H1N1 vaccine followed like-minded people. The public-health message is that if anti-vaccination communities online translate to real-world geographic pockets, it creates a greater risk of local outbreaks.
“By definition, herd immunity only works if unvaccinated, unprotected individuals are distributed sparsely throughout the population, buffered from the disease by vaccinated individuals,” the researcher said. “Unfortunately, the data from Twitter seem to indicate that the buffer of protection cannot be counted on if these clusters exist in real, geographical space.”
Because Twitter users often include a location in their profiles, the researcher could correlate tweets with Centers for Disease Control and Prevention data to determine how vaccination attitudes correlated with estimated vaccination rates. The highest positive-sentiment users were from New England, which had the highest H1N1 vaccination rate.
The method could be used to guide public-health initiatives, the researcher said. Targeted campaigns could be designed according to which region needs more prevention education, as well as predict how many doses of a vaccine will be required in a particular area.
The researcher plans to use his unique social-media analysis to study other diseases, such as obesity, hypertension, and heart disease. It’s already known that obese people are more likely to be associated with one another in real life.
John Snow would have been proud. Recall that he was the first epidemiologist, who in 1854 traced cholera in London back to a single source, a water pump used by the populace. He hypothesized that a single pump was responsible for more than 500 deaths by drawing a map that compared the locations of cholera deaths and the locations of water pumps throughout the city (perhaps making him the first user of Foursquare. Snow had to talk to people directly to gather his data, and would have been amazed at how advances in communication are helping modern scientists track disease.
*This blog post was originally published at ACP Internist*