Social Media and Wellbeing – or: Can your iPhone tell if you´re depressed?

For a moment, please imagine that you have had an episode of depression at an earlier point in your life. Obviously, you would not want anything of that to return anytime soon. Unfortunately, that´s just not an easy thing to do. Relapse rates for depression are rather high – thus, there is a considerable probability of experiencing at least one other episode once who been there for the first time.

Now imagine there´s someone that could give you an early warning. Someone that would be able to detect and interpret all those little behavioral signals that typically occur when a person slips in into depression. This someone would also automatically notify your doctor so she can catch up with you. Sounds like a dream of the future? Well, maybe not. What if your smartphone could understand what you are saying all day long – and then infer from your words that you are currently talking like somebody who is depressed?

Enter several researchers from the University of Pennsylvania, of whom one is Johannes Eichstaedt (who was in the MAPP program three years ago). They analyzed 700 million words and phrases collected from Facebook messages of 75,000 volunteers, who also took standard personality tests. What they´ve found is pretty amazing: the usage of certain (groups of) words on Facebook can be highly predictive of certain aspects of our personality, but also other variables like gender or age. So while there are a lot words that basically everybody uses to the same extent, there are specific words or sentences that on average tend to occur more often when, e.g., you´re a woman (as opposed to a man), or 35 years old (as opposed to 15), or extroverted (as opposed to introverted), or displaying high levels of Neuroticism (as opposed to Emotional Stability).

journal.pone.0073791.g006

Have a look at the image (click to enlarge), especially at the grey, blue, and red wordles at the center of the word clouds (those that are surrounded by the greenish ones). They can tell us something on the language(s) of a) extroversion, b) introversion, c) neuroticism, and d) emotional stability. The size of each word will tell you something on the predictive power pertaining to the variable in question.* By way of example, the use of the word ‘internet’ is a better predictor of being introverted than the use of ‘comic’. Additionally, the color will tell you how often that word is used (relatively; grey = not that often; blue: often; red = very often).

Now isn´t that cool?  But…you might ask: So what?

The ‘So what?’ leads us back to beginning of this post: if there is a typical ‘language of neuroticism’, there might also be a typical ‘language of depression’ – since displaying, e.g., a high level of neuroticism is correlated with the prevalence of depression. Or there might be a typical ‘language of mania’, or a ‘language of schizophrenia’ etc.

Now suppose there were an app on your smartphone that – at certain intervals over the day – switched on and took sound files of whatever you´re doing at a specific moment. It would surely pick up a lot of your conversations. By way of speech recognition (and prior, being fed with the algorithms that the abovementioned research is based on), your smartphone could detect if, over the course of time, your use of language changes from a ‘language of (relative) mental health’ to a ‘language of relative mental illness (perhaps, the app could also analyze whatever you´ve written on Facebook, Twitter and e-mails on a specific day). And if that were the case, the app would report this change back to you (or your doctor) as a means of early recognition. Wouldn’t that be really, really beneficial to a lot of people?

Now to date, this is a dream of future. But all the ingredients are there!

If you would like to learn more on this research, please click here for the original research paper. Also, there is a lot of cool stuff coming up in the near future – so you might want to check out the website of the World Well-Being Project.

 

*Please note that this is correlational research – so it is not appropriate to make any causal inferences. For instance, frequently using the word ‘party’ will not make you more extroverted. Rather, it can be likened to a ‘side effect’ of already being extroverted.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s