New research uses over half a million Facebook status updates to predict depression diagnoses in people at risk.
Worldwide, the World Health Organization (WHO) estimate that unipolar depressive disorders will be “the leading cause of the global burden of disease” by 2030.
New research aims to help create better screening and diagnostic tools for depression by using the information provided by social media.
Researchers, jointly led by Johannes Eichstaedt, founding research scientist at the World Well-Being Project (WWBP) in Philadelphia, PA, and H. Andrew Schwartz, a principal investigator of the WWBP, used an algorithm to analyze social media data from consenting users and picked out linguistic cues that might predict depression.
The team published their findings in the journal Proceedings of the National Academy of Sciences. Johannes Eichstaedt is the first author of the paper.
Eichstaedt and colleagues analyzed data from almost 1,200 people who agreed to provide their Facebook status updates and their electronic medical records. Of these participants, only 114 had a history of depression.
Study co-author Raina Merchant says, “For this project, all individuals [have] consented, no data is collected from their network, the data is anonymized, and the strictest levels of privacy and security are adhered to.”
Then, for every person who had received a diagnosis of depression in their lives, the researchers matched another five controls who had not. In this way, the researchers matched 683 people.
The scientists fed the information into an algorithm. In total, Eichstaedt and colleagues analyzed 524,292 Facebook status updates from both people who had a history of depression and from those who did not.
The updates were collected from the years leading up to a diagnosis of depression and for a similar period for depression-free participants.
By modeling conversations on 200 topics, the researchers determined a range of so-called depression-associated language markers, which depicted emotional and cognitive cues, including “sadness, loneliness, hostility, rumination, and increased self-reference” — that is an increased use of first-person pronouns, such as “I” or “me.”
Eichstaedt and team proceeded to examine how often people with depression used these markers, compared with controls.
The researchers found that the linguistic markers could predict depression with “significant” accuracy up to 3 months before the person receives a formal diagnosis.
“Unobtrusive depression assessment through social media of consenting individuals may become feasible as a scalable complement to existing screening and monitoring procedures,” conclude the authors.
The study’s first author also comments on the findings, saying, “The hope is that one day, these screening systems can be integrated into systems of care.”
“This tool raises yellow flags; eventually the hope is that you could directly funnel people it identifies into scalable treatment modalities,” Eichstaedt continues.
The researcher goes on to compare their social media algorithm with a DNA analysis. “Social media data contain markers akin to the genome,” Eichstaedt says.
“With surprisingly similar methods to those used in genomics, we can comb social media data to find these markers. Depression appears to be something quite detectable in this way; it really changes people’s use of social media in a way that something like skin disease or diabetes doesn’t.”
“[Social media] may turn out to be an important tool for diagnosing, monitoring, and eventually treating it. Here, we’ve shown that it can be used with clinical records, a step toward improving mental health with social media.”
H. Andrew Schwartz