Utility of Emotion and Sentiment Patterns for Early Screening of Depression

Tamang, Pemba; Xiao, Lu

Permalink

https://hdl.handle.net/2142/122831 Copy

Description

Title

Utility of Emotion and Sentiment Patterns for Early Screening of Depression

Author(s)

• Tamang, Pemba
• Xiao, Lu

Issue Date

2024-03-20

Keyword(s)

• Natural Language Processing
• Depression Screening
• Mental Health

Abstract

Early detection of depression reduces an immense amount of unnecessary mental, physical, and financial suffering for the individual. It also increases the efficiency of limited resources of the mental health system. The devel-opment of objective screening tools for early detection of depression has be-come increasingly popular because it is one of the primary ways to reduce discrepancies to the lack of mental health access. Interested in the potential of using sentiment polarity and emotion patterns in user-generated social media content for early detection of a user's depression state, we analyzed four publicly available labeled text datasets using pre-trained transformer models. All four datasets mainly consisted of social media Facebook, Twitter and Reddit posts labeled as depressive (text) or non-depressive (text). Our analysis shows that negative sentiment is significantly higher and positive sentiment significantly lower in all texts labeled as depressive across all the datasets. In addition, fear followed by sadness have the strongest positive correlations with texts labeled as depressive, whereas joy, neutral and sur-prise have the strongest negative correlations. These findings indicate that emotion and sentiment are of great utility in developing early detection tools that are based on user-generated social media content as they exhibit signifi-cant and observable patterns in the content.

Publisher

iSchools

Series/Report Name or Number

iConference 2024 Proceedings

Type of Resource

Other

Language

eng

Handle URL

https://hdl.handle.net/2142/122831

Copyright and License Information

Copyright 2024 is held by Pemba Tamang and Lu Xiao. Copyright permissions, when appropriate, must be obtained directly from the authors.

Owning Collections