Tracking Disease With Twitter: Social Media Epidemiology
In health, as in many other sectors, social media such as Twitter are attracting lots of interest. The possible applications for the various technologies are limited only by the imagination. For example, could Twitter provide better post-market drug surveillance than the Food and Drug Administration (FDA)? Its possible. Its fun to think about. But where is the evidence? To date, health research on social media has been relatively sparse. But that is beginning to change.
One of the most interesting analyses has just been published in the online, open-access publication PLoS. The study looks at Twitter traffic during last years H1N1 swine flu epidemic, and compares the volume of tweets about H1N1 with disease incidence and news coverage. After reviewing some 2 million tweets between May and December 2009, they found:
- H1N1-related tweets grew from 8.8% to 40.5% of all tweets.
-The shift from “swine flu” to “H1N1” occurred in a linear fashion following the WHO’s adoption and dissemination of the more clinical term.
-News websites were the most popular tweeted info source, providing 23% of the content.
-Government sites were tweeted 1.5% of the time.
-4.5% of tweets were classified as “misinformation.”
-The volume of H1N1 related tweets also peaked at times when major news stories about the epidemic were released, such as when the WHO raised their pandemic status to level 6.
-One of the sharpest increases in Twitter traffic was the announcement that Harry Potter actor Rupert Grint had swine flu.
-Overall, the volume of tweets about H1N1 “correlated well with disease incidence.”
The researchers hypothesize on the notion of whether their “infoveillance” research suggests a social media based cousin of epidemiology that they call “infodemiology.” Fun stuff.
So does this mean that we should stop looking at patients, populations and clinics to assess the state of the next global pandemic, and instead follow Twitter trends? Errrrr, probably not. Twitter might be a better reflection of what’s in the news, rather than what’s in the blood stream, since it is essentially an amplifier of already reported observations. But the fact that traffic was correlated, in retrospective analysis, to overall disease trends is provocative. Are we entering an age of crowd-sourced global health?
The media is also young, and usage is developing. If folks begin to tweet about strange health happening around them, and link them with hashtags, we might eventually get to the point of using social media as an epidemiological tool. If nothing else, it could serve as an early warning trigger to public health officials that someone needs to have a look to see if something is in fact afoot.
Naturally this doesn’t just apply to health, but could also foretell of political repression and human rights concerns, looming famines, strange weather phenomenon. The trick is how do we coordinate and monitor in an environment that thrives specifically because it isn’t coordinated and monitored?
What’s your take, will Twitter ever become a meaningful way of tracking disease?
Photo credit: PLoS and twitter.com