Depressive language codes About facebook entries Diagnose depression

Depressed people can be recognized by their language on the internet

Depression has long become a common disease. As the German Depressionshilfe Foundation reports, about one in four women and one in eight men suffer at least one depression during their lifetime. But by no means all sufferers receive the help they need to defeat the disease. This is not least because many people do not acknowledge their illness or actively seek help. American researchers have now developed an algorithm that scans entries from social media to filter out those who either suffer from depression or are at increased risk of developing it.

American researchers from the University of Pennsylvania and Stony Brook University relied on recent research that decoded a kind of language code from depressives. From this language code, the science team programmed an algorithm that scans entries on social networks, such as Facebook, to detect individual users' warning signs of the onset or rapid development of depression. The study results were recently published in the renowned journal "Proceedings of the National Academy of Science" (PNAS).

People who suffer from depression change the way they communicate over the Internet. A new algorithm can detect these changes and identify depressive individuals. (Image: Photographee.eu/fotolia.com)

Recognize depression before it arises

According to the German Depressionshilfe Foundation, around 5.3 million people in Germany are affected by depression. The newly developed algorithm could help accurately predict future depression before a medical diagnosis was made. More people could be helped, which in many cases is needed to defeat a depressive illness.

The language of depression

Several studies have confirmed that depressives use a striking language. This is characterized by negative adjectives such as lonely, sad or unhappy as well as words such as tears, pain, feelings, loneliness and hostility. In addition, depressive persons more often use the ego pronoun "I," but much fewer second or third person pronouns like you, he, she, or it.

How social media can help with the diagnosis of depression

"What people write on social media captures an aspect of life that is difficult to access in medicine and research," senior author H. Andrew Schwartz reports in a press release on the study's findings. The research team wants to use this information as a disease marker to uncover depression, anxiety disorders and post-traumatic stress disorder.

Six years of research

From a six-year language analysis, researchers developed a program that can detect and predict depression among social media users. "Depression seems to be provable in this way, since those affected change the use of social media in a very specific way," explains Schwartz. For a skin disease or diabetes, this would not be the case.

Similarly good results as a screening test

The researchers analyzed the data from the social media of 1200 participants. Of those, 114 people suffered from depression. The algorithm should now independently recognize the depressed person. In addition he scoured more than 500,000 entries. In fact, the program managed to detect depression with similar reliability as current screening tests.

The language changes within months

As a control, the researchers evaluated entries from depressive people who were older than six months. Here, in many cases, the algorithm did not detect depression, suggesting that the language has actually changed.

An unobtrusive depression test?

The University of Pennsylvania scientist Johannes Eichstaedt, who is involved in the study, sees long-term potential in the algorithm. He could be used as an unobtrusive depression test without those affected have to answer unpleasant questions. He hopes that this program will one day be integrated into the health system. (Vb)