Written by Nancy Lapid
(Reuters) – Using artificial intelligence to analyze social media may detect signs of depression in white Americans but not in black people, researchers show. According to a study highlighting the risks of training AI models for healthcare-related tasks without data from ethnic groups.
Researchers reported that the AI model used in the study was more than three times less likely to predict depression when applied to Black people using Metaplatforms’ Facebook than when applied to white people.
“Race appears to have been particularly neglected in studies of language-based assessment of mental illness,” the authors of the US study said in a report published in PNAS, the Proceedings of the National Academy of Sciences.
Previous research on social media posts has found that people who frequently use certain categories of words, such as first-person pronouns such as “me,” “me,” and “my,” as well as self-deprecating words, are more likely to experience depression. It has been shown that there is a high risk of
In the new study, researchers used an “off-the-shelf” AI tool to analyze language in posts from 868 volunteers, including an equal number of black and white adults who shared other characteristics such as age and gender. did.
All participants also completed a validated questionnaire used by health care providers to screen for depression.
Sharath Chandra Gangtuk of the Insight Center, a co-author of the study, said that the use of “private talk” and self-centered attention, self-deprecation, self-criticism, and feeling like an outsider are He said that it was only associated with depression. Accomplishments at Penn Medicine.
“We were surprised that the associations between these languages found in many previous studies did not hold true across the board,” Gangtuk said.
Gangtuk acknowledged that while social media data cannot be used to diagnose patients with depression, it could be used to assess risk for individuals or groups.
Previous research by his team analyzed language in social media posts to assess community mental health during the COVID-19 pandemic.
Brenda Curtis, director of the National Institute on Drug Abuse at the National Institutes of Health, found that among people with substance abuse disorders, social media language indicative of depression can provide insight into the likelihood of treatment discontinuation and relapse. He said that it has been shown that. the study.
(Reporting by Nancy Lapid; Editing by Michelle Gershberg and Bill Berkrot)
