Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?

Currently accepted at: Online Journal of Public Health Informatics

Date Submitted: Jun 24, 2026
Open Peer Review Period: Jul 3, 2026 - Aug 28, 2026
Date Accepted: Nov 10, 2026
(closed for review but you can still tweet)

This paper has been accepted and is currently in production.

It will appear shortly on 10.2196/105426

The final accepted version (not copyedited yet) is in this tab.

Artificial Intelligence in Reddit Mental Health Discourse: Scoping Review of Analytical and Conversational Applications

  • Aleksandra Martinsone; 
  • Kristīne Martinsone; 
  • Aleksejs Ruža; 
  • Anita Pipere

ABSTRACT

Background:

Artificial intelligence (AI) has rapidly expanded the methodological possibilities for analyzing mental health discourse in online environments. Reddit has emerged as a valuable source of user-generated mental health discussions because of its large, anonymous, and topic-specific communities. At the same time, conversational AI has become both a subject of discussion and a resource that some users describe as supportive within these communities.

Objective:

This scoping review with a narrowly defined scope with a narrowly defined scope examined how AI has been used to analyse mental health discussions on Reddit and how Reddit users experience and evaluate conversational AI for mental health support. It addressed four review questions concerning AI methods, the operationalisation of mental health constructs, user experiences with conversational AI, and perceived benefits, risks, and limitations.

Methods:

A structured search of a single database (Scopus) identified 925 records, and 1 additional report was identified through citation tracking. After title and abstract screening of 924 records and full-text assessment of 251 reports, 18 studies published between 2020 and 2025 met the inclusion criteria. The included studies were classified into three analytically distinct streams: Analytical AI (n = 15), Conversational AI – User Experience (n = 2), and Conversational AI – Technical (n = 1). Contextual qualitative studies without an AI component were treated as background literature and are not counted in the evidence base. Data were extracted using a structured charting framework and synthesized narratively. Reviewer-generated interpretation, guided by the sensitizing concepts of construct validity, contextual dependence, and epistemological assumptions, was labeled separately from the findings reported by the included studies.

Results:

Analytical AI studies predominantly applied supervised machine learning, transformer-based models, and topic modeling to analyze linguistic and thematic patterns associated with psychological distress. These studies frequently relied on proxy labels and internal performance metrics, with limited attention to construct validity and contextual interpretation. The two user-experience studies provided early signals that some Reddit users perceive conversational AI as accessible and non-judgmental, while also expressing concerns about inaccurate advice, excessive affirmation, privacy, and its limited ability to replace human therapy; these are user-reported perceptions rather than clinical outcomes. The single technical study demonstrated residual hallucination risk even when clinical process knowledge was incorporated. No included studies provided controlled evaluations of the safety or efficacy of conversational AI for mental health purposes.

Conclusions:

AI provides valuable tools for analyzing large-scale mental health discourse on Reddit, although the analytical evidence rests on proxy labels and overlapping datasets. Evidence on conversational AI is limited to early signals from two user-experience studies, which describe perceived support and user-reported concerns rather than demonstrated therapeutic outcomes. None of the included studies provided controlled evidence of the safety or effectiveness of conversational AI for mental health care. Important methodological and ethical challenges remain regarding construct validity, contextual interpretation, and responsible implementation. Future research should prioritize theoretically grounded construct operationalization, greater methodological transparency, and rigorous real-world evaluation of conversational AI. Bringing analytical and conversational AI into a single evidence framework highlights both the opportunities and the unresolved challenges of AI use in Reddit mental health research.


 Citation

Please cite as:

Martinsone A, Martinsone K, Ruža A, Pipere A

Artificial Intelligence in Reddit Mental Health Discourse: Scoping Review of Analytical and Conversational Applications

Online Journal of Public Health Informatics. 10/11/2026:105426 (forthcoming/in press)

DOI: 10.2196/105426

URL: https://preprints.jmir.org/preprint/105426

Download PDF


Request queued. Please wait while the file is being generated. It may take some time.

© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.