Depression is often not diagnosed. Researchers are turning to Facebook to change that. | Lifestyles

Depression affects more than 16 million Americans a year, but less than half of them seek treatment. Researchers are now turning to social media to narrow this gap and give doctors another way to find out who is at risk.

A study published in the Proceedings of the National Academy of Sciences suggests that Facebook's language analysis publications can predict whether a user is depressed three months before the person receives a medical diagnosis.

The work is still in its early stages, researchers at the University of Pennsylvania and Stony Brook University warned. The study was based on a group of less than 700 users and the predictive model is only moderately accurate. But this approach could be promising for the future, they said.

"Depression is a very debilitating disease and we have treatments that can help people," said Raina Merchant, one of the study's authors and director of the Penn Health Medicine Center for the digital health. "We want to think of new ways to get resources and people's identification for depression sooner."

The researchers recruited study participants in the emergency room of a hospital, requesting permission to access their electronic medical records and their Facebook history. The researchers found five people who had not diagnosed depression in their medical records, creating a sample reflecting the rates of depression in the national population.

By examining more than 500,000 Facebook posts in both groups, researchers determined which words, length of publication, frequency of publication, and date of publication were most often associated with a diagnosis of depression. They found that depressed people used the words "me, me, and me" as well as words such as "hurt, fatigued, and hospitalized" more often than others in the months prior to their diagnosis. Using such indicators, they developed a computer model to predict which people would be diagnosed with depression with a precision comparable to commonly used clinical surveys.

The model worked better with the Facebook data of the three months preceding the diagnosis of depression by a participant. When longer periods of Facebook data were included, the model became less accurate.

"We are about to try to understand how these data are sometimes words of common salvation, but they can sometimes give us insight into the health of individuals and communities," Merchant said.

The symptoms of depression manifest themselves differently according to race, gender and age and may be affected by other diseases, making diagnosis difficult. Most screening tools are based on the fact that people accurately report their own symptoms and answer questions from the survey, which can be interpreted differently depending on the cultural background and language skills of an individual. nobody.

Primary care physicians can screen for depression, but their visits to patients are often short and spaced by several months, leaving the focus on crises and immediate concerns.

"With social media and other data, you can begin to fill those gaps," said Munmun De Choudhury, assistant professor at Georgia Tech's School of Interactive Computing. His previous research has shown that Twitter data can be used to predict which users will develop symptoms of depression.

In the future, if patients shared social media data with their doctors, this could create more personalized care, De Choudhury said. "How's their social life going, are they getting enough, a lot of these attributes that you can measure using social media," she said.

Social media data could also be used for public health, De Choudhury said. For example, disease control and prevention centers could identify communities most at risk of suicide by reviewing their online publications and then providing them with specific prevention measures.

Facebook and Google have begun to take steps in this direction. Facebook uses artificial intelligence to report publications indicating risks of self harm or suicide. From there, an employee can refer people to national suicide prevention resources. Google invites users looking for depression-related terms to answer a selection questionnaire.

It's encouraging to see these companies take on social responsibilities, De Choudhury said, but that can only be one aspect of mental health care. Predictive models built on social media are not yet very accurate. They are also built on small samples, which means that they may not work the same way in a large and diverse population.

"You should not use such an algorithm alone at any time," she said. It must be combined with traditional screening surveys for depression and clinical expertise.

The privacy issue is another reason to remain cautious with the use of social media for health care, Merchant said. "We should see this data as we process all the health data," she said. "This is the patient data." But this is a delicate premise given the recent high-profile data breaches, including an offense that has compromised millions of Facebook users.

Some also fear that social media will do more than reflect their mental health. Some studies have shown that people using more social media are more likely to be depressed or suffering from eating disorders. But other studies show that social media can be helpful in connecting people to resources and peer support.

More research is needed, Merchant said. "We need to better understand not only how it tells us about our health, but also how the use of technology affects our health."

___

(c) 2018 The Philadelphia Inquirer

Visit the Philadelphia Inquirer at www.philly.com

Distributed by Tribune Content Agency, LLC.

_____

PHOTO (for help with images, contact 312-222-4194):