Depression Symptoms and Relational Patterns in 187k ChatGPT Histories

2026-07-06

AI Summary

This study presents the first large-scale empirical analysis of how individuals exhibiting depressive symptoms utilize large language models (LLMs) for informal psychological support. Drawing on 187,093 ChatGPT conversations, the research integrates PHQ-8 symptom severity stratification, natural language processing, and temporal behavioral modeling to reveal that users with higher depressive symptomatology engage more frequently in discussions about mental health, loneliness, and interpersonal relationships, and exhibit significantly greater use of first-person pronouns and absolutist language. Although machine learning models can partially identify such users (AUROC = 0.591), their predictive performance remains insufficient for clinical screening purposes. The findings highlight the potential of LLMs as a private, continuous, and accessible infrastructure for psychological support while cautioning against the use of such interaction data for diagnostic applications.

Abstract

Large language models are increasingly used as private, always-available conversational systems, but little is known about how people with depressive symptoms use them. Building on CSCW work on disclosure and peer support, we examine ChatGPT as an emerging informal support infrastructure: private, persistent, responsive, and available outside ordinary hours. We analyze 187,093 ChatGPT conversations from 766 participants who completed the PHQ-8, comparing those below the moderate-symptom threshold (score of 10) with those at or above it. Higher-PHQ participants used ChatGPT more for mental-health, interpersonal, loneliness, self-focused, and support-seeking conversations, with pronounced late-night and recurring month-level patterns. Their language contained more first-person singular pronouns and absolutist terms. They more often engaged ChatGPT in high-disclosure contexts, but professional redirection was not higher. Language-based prediction was modest and insufficient for screening (AUROC 0.591). We argue these histories should not be treated as clinical screening data but as evidence LLMs are increasingly used as informal support infrastructure.

Problem

• depression symptoms
• large language models
• informal support
• conversational patterns
• mental health disclosure

Innovation

• large language models
• depression symptoms
• informal support infrastructure
• disclosure patterns
• mental health