Sunny Rai, PhD: “I” Language Markers Do Not Detect Depression in Black Individuals

News
Video

In an interview with HCPLive, Sunny Rai, PhD, shed light on a new study regarding how an AI model failed to detect depression in Black individuals.

Social media posts filled with “I,” “Me,” and “My” can detect depression, as previous research suggested, but not for Black people, a new study found.1

The new data revealed White individuals are more likely to use the “I” personal pronoun in Facebook posts—the only social media platform evaluated in the study—which may convey a negative emotion, such as self-deprecating terms or language expressing outsider feelings.

The findings came from an AI model that predicted depression using language in Black and White participants. The model analyzed ≥ 800 Facebook posts, half created by White individuals, and the other half from Black individuals. The model looked for “predictive words”—the personal pronouns—and found this mainly applied to White individuals. Due to this, the model could not effectively detect depression in Black people.

Investigators hypothesized the model worked poorer on Black individuals, despite the AI program being fed an equal number of posts from Black and White individuals, because more data is needed to learn the depression patterns in Black individuals. Another possible explanation could be that Black people do not show markers of depression on social media due to perceived stigma.2

In an interview with HCPLive, lead investigator Sunny Rai, PhD, from the University of Pennsylvania, highlighted the biggest takeaway from the study, how the findings may influence the approach to depression screening, and ways AI models could be improved.

“Surprisingly, we found that the use of first-person singular pronoun shows no correlation with [depression] among black participants,” Rai said. “Now, this has multiple implications because these words are known to reflect self-focused thinking or self-image perspective and are also used as indicators of depression. So, this means that probably these markers do not generalize well with other races.”

Rai explained the findings indicate depression might not be easy to capture from written language for all races, having different language markers. She said we need to understand more about how language varies across races and other demographic variables to improve the models.

“Our results raise concern that certain psychological processes thought to predict or maintain depression may be less relevant or even irrelevant to populations historically excluded from psychological research,” Rai said. “And that includes black individuals.”

References

  1. Rai S, Stade EC, Giorgi S, et al. Key language markers of depression on social media depend on race.Proc Natl Acad Sci USA. 2024;121(14):e2319837121. doi:10.1073/pnas.2319837121
  2. Depression in Black People Goes Unnoticed by AI Models Analyzing Language in Social Media Posts. News Wise. March 26, 2024. https://www.newswise.com/articles/depression-in-black-people-goes-unnoticed-by-ai-models-analyzing-language-in-social-media-posts. Accessed April 8, 2024.


Related Videos
Vlado Perkovic, MBBS, PhD | Credit: George Institute of Global Health
Elizabeth Aby, MD | Credit: Minnesota Health Fairview
Video 3 - "Insights Gleaned from Asthma Research for COPD"
Video 3 - "Insights Gleaned from Asthma Research for COPD"
Video 3 - "HIV Treatment: Discussing Adverse Events with Patients"
Prashant Singh, MD | Credit: University of Michigan
Sean Adrean, MD: Impact of Baseline VA on Aflibercept 8 mg Outcomes in DME | Image Credit: Linkedin
Video 3 - "Key Clinical Considerations in HIV Treatment Decisions"
Video 2 - "Lessons from EXPLORER-HCM: Unveiling CMIs' Potential in oHCM Treatment "
© 2024 MJH Life Sciences

All rights reserved.