Psychological and Behavioral Insights From Social Media Users: Natural Language Processing-Based Quantitative Study...

Area of Science:

• Computational linguistics
• Mental health informatics
• Social network analysis

Background:

• Depression affects millions globally, with traditional detection methods being time-consuming and potentially inefficient.
• Social media offers a new, unbiased perspective on language and behavioral patterns for mental illness understanding.

Purpose of the Study:

• To develop and evaluate a language framework integrating psychological patterns, context, and social interactions.
• To enhance machine learning-based depression detection at the user level using natural language processing (NLP).

Main Methods:

• Extracted language patterns (affective, personality) and social interaction features from social media posts using NLP.
• Developed a framework combining psychological and social influence features.
• Evaluated framework performance using machine learning on 1047 users' social media data and questionnaire scores.

Main Results:

• The framework achieved 77% accuracy and 80% precision using all influence features.
• Models incorporating affective and social influence features showed strong performance (81% precision, 79% F1-score).
• Outperformed traditional baselines by an average of 6% in accuracy and 10% in F1-score.

Conclusions:

• The developed framework enables accurate and effective depression detection.
• Offers practical applications for accelerating diagnoses, improving predictions, and facilitating early mental health interventions.
• Highlights the potential of social media data in mental health assessment.