Twitter has often been looked to as a way to improve and monitor public health, from researchers uncovering vaccine trends to tracking flu outbreaks. Now, a new research team has developed a method for tracking HIV outbreaks and drug use behavior using tweets.
Published in Preventative Medicine by UCLA researchers, this latest study examined the content of tweets to determine where potential HIV outbreaks might occur.
As PsychCentral describes, researchers collected over 550 million tweets between May 26 and December 9 2012. They explored this data using an algorithm to pull out words associated with drug use and risky sexual behavior, such as “sex” and “get high.” A total of 8,538 tweets containing reference to sexually risky behavior and 1,342 tweets referring to drug use were identified.
They then used geo-location to pinpoint where these tweets were sent from, and compared this data to data about reported HIV cases to see if there was overlap.
The study found the largest number of HIV-risk related tweets came from the District of Columbia, Delaware, Louisiana and South Carolina, on a per capita basis.
The researchers found a significant relationship between the tweets indicating risky behavior and the counties with the highest number of reported HIV cases. Based on this, they concluded that Twitter and other forms of real-time data could possibly be used to predict when HIV outbreaks and drug use occurred.
Like other similar studies, this one used data from the past – one of its major challenges will be to apply the model to real-time tweets to effectively predict future HIV outbreaks.
(Medical staff using computer image via Shutterstock)