Scientists are trying to model our mental health based on our tweets

During the holiday season, ideally filled with family, food, and festivities, the topic of depression is often sidelined; even more of a taboo subject than usual. But research suggests it is one of our most persistent blights, ranked ninth in the world behind the major killers, such as heart disease, stroke, and HIV, according to Nature.

Now researchers from multiple disciplines, in both the public and private sectors, are working on various algorithms and approaches to measure a range of mental health trends via large volumes of online activity. Issues such as depression and seasonal anxiety disorder aren’t the first health trends to be investigated in this way – think Google Flu Trends, for instance – but they represent an entry point for researchers, one that most recently has been hailed by a team at Johns Hopkins reporting on techniques that could play a key role in measuring mental health metrics.

“This is like a microscope or a telescope, but it’s something new: a macroscope,” Nova Spivack, CEO of LA-based Bottlenose, told me. Spivack said his company currently measures 72 billion messages in real time every day on Twitter, Facebook, Tumblr, search engines, as well as on TV and the radio in the US, Canada, UK, and the Middle East. “Inhale as much as you can and allow the trends to bubble up,” he said; from there his team uses tools like topic detection and sentiment analysis to make sense of data for global 1000 companies both historically and in real time.

Depression analysis on social media Nova Spivack

Spivack is a fervent believer in the power of big data – “It’s amazing that you can do that, literally measure billions of opinions; it’s a feedback loop that never existed, and it’s going to transform society,” he said – but the science itself is still in its infancy, and tapping platforms like Twitter to tell us about trends has its limitations, some obvious and others less so.

One of the inherent problems with relying on the online expressions of social media users is that, while massive amounts of data can help reduce statistical bias, the demographic itself becomes overrepresented. “You can get all these rare phenomena, but you suffer from this other one that social media alone doesn’t capture to really understand the picture of mental health,” Johns Hopkins computer scientist Glen Coppersmith, who is part of the team working on measuring mental health trends via Twitter, told me.

Even if Twitter users perfectly represented the global population, researchers scanning for tweets that contain clues to mental health issues are still only getting data from those willing to tweet about them. Still, even though offline introverts escape notice, much is being gleaned from that data streaming in, and Spivack stresses that a lot of times the statistical anomalies tell us things we didn’t even know we were looking for.

At Johns Hopkins, researcher are drumming up new numbers on the prevalence of certain mental health illnesses by looking for direct tweets mentioning diagnoses as well as by looking for language cues directly linked to certain disorders. The idea is to share their findings with public health officials to improve on the often slow and costly method of collecting data through surveys – though Coppersmith stresses that because it is data in aggregate that is of interest, no users are singled out or identified in any way.

“There’s a whole class of analyses that could be done that hasn’t been done,” he said. “The real key is we now have a lot of data that’s relevant, so the next thing we’re looking at is not a single diagnosis but being able to look at multiple diagnoses and multiple people who have been treated as separate diagnoses and look at how the language of these may be related. I’m very excited about it.”

Many other topics are also being explored on Twitter and beyond. Researchers are looking at suicidal thoughts, religious beliefs, global mood patterns, and more, and Twitter now sells its data to companies interested in niche trends, and turns a pretty profit in the process.

But for researchers like Coppersmith, the larger goal is to better our understanding of ourselves in the aggregate, and ideally improve our social infrastructures, too. And while we’re just beginning to explore the power of the “macroscope,” as Spivack puts it, Coppersmith says mental health provides a unique and often personal opportunity: “Mental Health has touched each one of our lives in some way: a friend, a relative, a colleague, or one’s self. When we saw the opportunity to try to make a dent in understanding some of these problems, it wasn’t hard to convince others to help out.”

Featured image courtesy of Thinkstock/Chalabala