Artificial Intelligence Uses Facebook Data to Predict Mental Illness

A study that used AI to detect mood disorders among Facebook users.

FREMONT, CA: It's easy to do bad things with Facebook data. From targeting bizarrely specific T-shirts to manipulating the electorate, the questionable goals to which the social media giant can be attributed are numerous. But there are also people who are trying to use Facebook for a good purpose - or at least to improve the diagnosis of mental illness. On December 3, a group of researchers reported that they had successfully predicted a psychiatric diagnosis based on Facebook data - using messages sent up to 18 months before the user received the official diagnosis.

The team worked with 223 volunteers, all of whom gave researchers access to their personal Facebook messages. Researchers use artificial intelligence algorithms, using attributes extracted from these messages and Facebook photos posted by each participant to predict whether they suffer from mood disorders (such as bipolar disorder or depression), schizophrenia spectrum disorder or not Mental health problems. According to their research results, foul language usually indicates mental illness, and perceptual words (such as seeing, feeling, hearing) and words related to negative emotions indicate schizophrenia. In the photo, more blue is related to mood disorders.

To assess the success of their algorithm, the researchers used a common metric in artificial intelligence to measure the trade-off between false positives and false negatives. As the algorithm classifies more and more participants as positive (for example, with schizophrenia spectrum disorder), it will miss fewer participants who really have schizophrenia (low false negative rate), but It classifies certain healthy participants as suffering from schizophrenia (high false positive rate). A perfect algorithm cannot have false positives and false negatives at the same time; such an algorithm will earn 1 point. The random guess algorithm will get 0.5 points. According to the specific predictions they asked the algorithm to make, the research team scored between 0.65 and 0.77. Even if researchers limit themselves to information more than a year before the subject’s diagnosis, the predictions they make may be much better than the results expected by chance.

According to H. Andrew Schwartz, associate professor of computer science at Stony Brook University, who was not involved in the study, these results are comparable to those obtained by PHQ-9, the standard 10-question questionnaire used to screen for depression. . This result creates the possibility that Facebook data could be used for mental illness screening - potentially long before the patient would otherwise have received a diagnosis.

Birnbaum is far from the first researcher to use social media data to predict the presence of mental illness. Previously, researchers used Facebook status, tweets, and Reddit posts to identify diagnoses ranging from depression to attention deficit hyperactivity disorder. But he and his team created a new situation by working directly with patients with existing psychiatric diagnoses. Generally speaking, other researchers who cannot make a clinically definitive diagnosis-they have used the words of the subjects to make a diagnosis, ask them to make a self-diagnosis, or have them receive a questionnaire like PHQ-9 instead of diagnosis. In contrast, everyone in Birnbaum's study received a formal diagnosis from a psychiatrist. And because researchers have clear dates for making these diagnoses, they can try to make predictions based on the messages sent by patients before they learned about their mental illness.

Researchers have a long way to go before designing these algorithms and figuring out how to implement them ethically. However, Birnbaum hopes that social media data can become a normal part of psychiatric practice in the next five to ten years.