Your posts on Facebook can reveal if you are depressed

2.2 billion assets of Facebook users use the platform to share all kinds of things: commitments. Group plans. Political misinformation. Cat pictures. But as reported by researchers this week, the words you post in your status updates could also contain hidden information about your mental health.

In the research described in this week's issue of Proceedings of the National Academy of SciencesScientists analyzed the language of Facebook status updates of study participants to predict future diagnoses of depression. The researchers say their technique could lead to a screening tool that identifies people in need of mental health support and a formal diagnosis, while raising serious questions about privacy in health.

If this survey trail sounds familiar to you, you can not imagine anything: scientists have been studying for years the association between Facebook and the mental state of its users, often without the consent of the people examined. At the beginning of this decade, scientists from Facebook and Cornell led an infamous [emotional contagion](http://www.pnas.org/content/111/24/8788), a study that unknowingly targeted the mood and relationships of more than half a million users from Facebook. More recently, Cambridge Analytica has used poorly acquired data. According to him, some 87 million Facebook users have developed personality profiles, which would allow marketers and political campaigns to broadcast more effective ads.

But many scientists continue to use sophisticated search methods to access Facebook's data. For example: by asking study participants to give their consent, log into their accounts and share their data – all in person – to provide unique access to that data. The overheads are huge; It may take years to build up a sufficient sample of population using the recruitment to the study. However, this effort may be worthwhile for social scientists, who often consider that Facebook's many user information is the most important data repository in the history of their field.

"We understand more and more that what people are doing online is a form of behavior that we can read with the help of machine learning algorithms, in the same way that we can read "Any other type of data in the world," says psychologist Johannes Eichstaedt, first author of the University of Pennsylvania. of the new PNAS researcher and co-founder of the World Well-Being Project, a research organization that studies how words used by social media reflect their psychological state.

To determine if a language on Facebook could predict a diagnosis of depression, Eichstaedt and his colleagues needed access to two types of personal data: social media accounts and electronic medical records. During a 26-month period, they contacted more than 11,000 patients in an emergency department in Philadelphia and asked them if they would be willing to share their EMRs and up to $ 1,000. seven years of status updates on Facebook.

Some 1200 patients accepted. Of these, 114 had medical records indicating a diagnosis of depression. Every year, about one in six Americans suffers from depression. To replicate this ratio in their final research population, the researchers compared each person diagnosed with depression to five others. This gave the researchers a final pool of 684 participants. Using more than half a million Facebook status updates on these people, the researchers determined the most frequently used words and phrases and developed an algorithm to identify what they call markers of language associated with depression.

They found that depressed people used the "I" language more (for example, singular pronouns in the first person) and words that reflected hostility and loneliness in the months preceding their clinical diagnosis. By training their algorithm to identify these linguistic patterns, the researchers were able to predict future depression diagnoses up to three months before they appeared in their medical records as a formal condition.

The researchers' observation that depressed people use the "I" language more often matches findings from previous studies, including those with social media use patterns related to depression. self-reported. But this is the first study to compare the language used by Facebook users to clinical diagnoses using medical record data. "This is an important advance," says Matthias Mehl, a psychology researcher at the University of Arizona, who is studying how the use of language can reflect the psychological state of life. A person, "but the predictions are still far from perfect". The probability of the algorithm detecting symptoms of real depression, he says, was greater than the probability of false alarm, but was far from sufficient to replace a formal diagnosis.

Eichstaedt agrees. "It would be irresponsible to take this tool and use it to say: you are depressed, you are not depressed," he says. What it is could to be appropriate is to find people who should follow with more formal – and often more costly – screening methods. He added that future studies will have to replicate his team's findings to more and more diverse populations (the participants in this study were mostly black women).

This assumes that people will be willing to dissect the language they use on social media in search of mental health signatures – an infernal assumption, in light of the many Facebook privacy scandals. . And even if people agree to share their personal information, Eichstaedt says that you can not unlock its true predictive power until you combine it with another form of data: heart rate, activity or sleep, for example, which are all registered. more and more by the activity trackers.

"A benevolent dictator would connect all these data streams and use them for the public good," said Eichstaedt. But the moral and ethical alignments of the largest companies in the technology sector are under more scrutiny than ever before. If a filtering tool like this is debuting on Facebook or any other social media platform, it's hard to imagine it happening anytime soon.

Biggest cable stories