Researchers track mental illness with Twitter

In a time when sharing thoughts and feelings in the form of a hashtag has become the norm, researchers at Johns Hopkins University have turned to Twitter to gather important information about mental illness.

The new technique has already been used to track flu cases, and researchers told using social media provides a timely and effective way to obtain mental health trends.

Measuring mental health statistics by traditional methods, like phone surveys, has been notoriously difficult and sometimes inaccurate due to the reluctance of sharing sensitive information with a stranger. But people feel more comfortable opening up to the seemingly faceless “Twittersphere.”

“Social media is like a journal, which has a long tradition of being used in
psychology research,” said study author Glen Coppersmith, a research scientist at Johns Hopkins University’s Center for Language and Speech Processing.

Researchers have been able to collect new data on post-traumatic stress disorder (PTSD), depression, bipolar disorder and seasonal affective disorder (SAD) by reviewing tweets from users who publicly mentioned their diagnosis and by looking for language cues linked to certain disorders.

The data collection, conducted throughout 2013, was completed in three phases. First, hundreds of millions of public tweets were searched for statements of self-diagnosis, such as “I was diagnosed with depression.” Then, algorithms based on natural language processing were built to find tweet patterns that indicated mental health issues.

“Some of these differences are obvious and make immediate sense, like depressed people talk about negative emotions more,” Coppersmith said. “Some differences are more subtle, like depressed people use ‘I’ more than their community control counterparts.”

Lastly, the algorithms were applied on the populations level, such as military and non-military. The findings did not disclose the names of people who publicly tweeted about their disorders.

Their analyses indicated there was a higher rate of PTSD in military installations that frequently deployed during the recent Iraq and Afghanistan conflicts. Additional data showed increased signs of depression in locations with higher unemployment rates.

“Using Twitter to get a fix on mental health cases could be very helpful to health practitioners and governmental officials who need to decide where counseling and other care is needed most,” said study author Mark Dredze, and an assistant research professor in the Johns Hopkins University Whiting School of Engineering’s Department of Computer Science.

Researchers said they’re not aiming to replace the long-standing survey methods of tracking mental illness trends, but hope the new Twitter technique could complement the existing process by getting similar results more quickly and at a lower cost.

“Our next step is to use our new tool to test new theories that have previously been un-testable because we lacked the data. Our methods create new research opportunities for population mental health research,” Dredze added.