Twitter knows when you’re depressed, according to research. Simply by looking over your Twitter feed, scientists can predict – with 70 percent accuracy – when you’re feeling sad and blue.
The online social networking service, Twitter, was created in 2006. It allows users to post 140-or-less characters per tweet. The micro-blog Twitter has over 500 million registered users who post millions of tweets per day.
According to the Los Angeles Times, the United States is number one as the country with the largest number of actively tweeting Twitter users at 24.3 percent. Based on available stats, we’re immediately followed by Japan with 9.3 percent.
Thus, with Twitter’s broadening global prevalence, investigators have begun using the social platform in order to further public health research.
Tyler McCormick of the Center for Statistics and the Social Sciences at the University of Washington explains in TIME, “Our attitude is that Twitter is the largest observational study of human behavior we’ve ever known, and we’re working very hard to take advantage of it.”
One avenue of research is using study models – artificial intelligence and algorithms – to scan your Twitter feed to determine if you’re at risk for depression.
Co-director of Microsoft Research Redmond, Eric Horvitz and his colleagues pondered if it was possible to build measures capable of calculating and detecting if someone is severely depressed using social media posts.
The research team developed a tweet scan model intended to predict depression in 476 Twitter users, 171 of whom were actually seriously depressed. They scanned profile histories going back a year – studying use of language, engagement and interaction, and key words regarding medications and mood. This equated to over 2.2 million tweets.
The depression prediction study was 70 percent accurate when comparing depressed users’ feeds with the non-depressed user samples. Investigators admit the process is still imperfect. Tweaks are still necessary in order to yield out the remaining 30 percent and adjust for false positives – the 10 percent of non-depressed Twitter users inaccurately gleaned from the scans.
One interesting note from the research: When compared to emotionally healthy users, seemingly depressed Twitter users tweet between the hours of 9 p.m. and 6 a.m.
[Image: Mike Bailey-Gates]