The DOI matches its Crossref record; 24 of the 24 references it cites were found in a registry or an index; 6 of the 12 sources its claims cite were checked for support.
Role of large language models in mental health research: an international survey of researchers' practices and perspectives
Open accessAbstract
BACKGROUND: Large language models (LLMs) offer significant potential to streamline research workflows and enhance productivity. However, limited data exist on the extent of their adoption within the mental health research community. OBJECTIVE: We examined how LLMs are being used in mental health research, the types of tasks they support, barriers to their adoption and broader attitudes towards their integration. METHODS: 714 mental health researchers from 42 countries and various career stages (from PhD student, to early career researcher, to Professor) completed a survey assessing LLM-related practices and perspectives. FINDINGS: 496 (69.5%) reported using LLMs to assist with research, with 94% indicating use of ChatGPT. The most common applications were for proofreading written work (69%) and refining or generating code (49%). LLM use was more prevalent among early career researchers. Common challenges reported by users included inaccurate responses (78%), ethical concerns (48%) and biased outputs (27%). However, many users indicated that LLMs improved efficiency (73%) and output quality (44%). Reasons for non-use were concerns with ethical issues (53%) and accuracy of outputs (50%). Most agreed that they wanted more training on responsible use (77%), that researchers should be required to disclose use of LLMs in manuscripts (79%) and that they were concerned about LLMs affecting how their work is evaluated (60%). CONCLUSION: While LLM use is widespread in mental health research, key barriers and implementation challenges remain. CLINICAL IMPLICATIONS: LLMs may streamline mental health research processes, but clear guidelines are needed to support their ethical and transparent use across the research lifecycle.
Overview
CoChat Summary
TL;DR
This international survey of 714 mental health researchers found that 69.5% had used LLMs in research, most commonly ChatGPT for proofreading and coding. Users reported efficiency gains alongside frequent concerns about inaccurate responses, ethics, bias, and the need for training and disclosure guidelines.
Key findings
- 496 of 714 researchers (69.5%) reported using LLMs for research; among users, 94.4% used ChatGPT, 69.0% used LLMs for proofreading or improving writing clarity, and 49.0% for generating or refining code.
- LLM use was more prevalent among early career researchers: fewer years of research experience (OR 0.67, 95% CI 0.53 to 0.84) and more prior AI experience (OR 3.73, 95% CI 2.84 to 4.89) were significantly associated with greater adoption.
- Among users, common challenges were inaccurate or misleading responses (78.6%), ethical concerns (45.8%), and biased outputs (27.0%); 72.6% said LLMs improved efficiency and 44.0% said they improved output quality.
- Among non-users, top reasons for non-use were ethical concerns about data privacy and plagiarism (53.2%) and concerns about accuracy and biases (50.0%); most respondents wanted more training on responsible use (76.7%), required disclosure of LLM use in manuscripts (79.8%), and were concerned about LLMs affecting how their work is evaluated (60.4%).
Methods
- Cross-sectional online survey delivered via Qualtrics to researchers whose self-identified primary focus was mental health research, recruited between 18 March 2025 and 5 April 2025.
- Approximately 4000 corresponding authors from Scimago “Psychiatry and Mental Health” and “Clinical Psychology” journals were invited; 749 responded (19% response rate), 35 were screened out, and 714 met inclusion criteria.
- Analyses included descriptive statistics and percentages, plus a multivariable logistic regression examining LLM adoption with age, gender, h-index, years’ experience, publication count, and prior AI experience as simultaneous predictors.
Limitations
- The paper states potential self-selection bias: researchers more interested in AI may have been more likely to respond, possibly overestimating LLM awareness, use, or acceptance.
- The paper states most respondents were from high-income countries, most commonly Australia (24%), the USA (26%), and the UK (10%), limiting generalisability to low- and middle-income countries.
- The paper states the survey was not formally piloted or cognitively tested, and its reliance on predefined response categories may have limited depth; evident from the cross-sectional design, causal relationships cannot be established.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
6 Valid · 0 Not valid · 2 Not checked · 30 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“a prior study demonstrated high hallucination rates when LLMs were used to generate references for human-performed systematic reviews, with hallucination rates reaching 39% for ChatGPT models and 91% for Gemini.20”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Discussion
What the sources say
- Supportedcites [20] Chelli, 2024, Hallucination Rates and Reference Accuracy of ChatGPT and Bard for Systematic Reviews: Comparative Analysis, J Med Internet Res source
The Abstract Results report hallucination rates of 39.6% for GPT-3.5, 28.6% for GPT-4, and 91.4% for Bard. The Abstract Objective identifies Bard as subsequently rebranded Gemini; the reported figures support the citing sentence’s approximate rates, with 39% referring to GPT-3.5 rather than both ChatGPT models. (LLM, conf=0.91)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“LLMs are prone to biased outputs due to the nature of their training data, which often reflects societal inequalities and fails to uniformly represent diverse demographic groups.21”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Discussion
What the sources say
- Partly supportedcites [21] Straw, 2020, Artificial Intelligence in mental health and the biases of language based models, PLoS One sourceExperimental — accuracy not yet measured
Section 2.3 says biased datasets can affect NLP models, and section 4.3 calls for data produced by balanced demographic groups. These passages support the concern about bias and representation, but the paper discusses NLP models in mental health rather than establishing the claim specifically for LLMs. (LLM, conf=0.84)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“their potential to support academic research remains underexplored despite being potentially more fruitful.14”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Background
What the sources say
- Partly supportedcites [14] Hua, 2025, From statistics to deep learning: Using large language models in psychiatric research, Int J Methods Psychiatr Res sourceExperimental — accuracy not yet measured
The Introduction says prior reviews focused largely on clinical settings and missed LLMs’ potential in broader psychiatric research. This supports the underexplored-potential point within psychiatry, but the cited passages do not establish that academic research generally is underexplored or that it is potentially more fruitful than clinical use. (LLM, conf=0.85)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“they have ethical concerns about LLM use in research in general (78.3%)”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Findings > Broader perspectives of LLM integration in mental health research
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“496 (69.5%) reported using LLMs to assist with research”
- Also stated in: § LLM use · plain
“When asked whether they had ever used LLMs to assist with their research, 496 (69.5%) participants responded ‘yes’ and 218 (30.5%) responded ‘no’.”
- Also stated in: § Discussion · hedged
“Nearly 70% of researchers reported using LLMs—most commonly ChatGPT—to assist with their academic research”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Abstract > Findings
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § LLM use · plain
12 more claims
Create a free account to see all 38 claims and to verify the 1 reference still open.
- Probably not a substantial claim. 7
- It may state two separate claims as one. 15
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, while most LLM adopters reported that these tools enhanced research efficiency and improved the quality of their work, many also expressed concerns about inaccuracies, misleading content, and biases in the responses generated by these models [4].”
From Influence of Topic Familiarity and Prompt Specificity on Citation Fabrication in Mental Health Research Using Large Language Models: Experimental Study (2025) · cites this paper as [4]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“We selected GPT-4o because it is the most recent and widely adopted LLM among academic researchers [4]”
From Influence of Topic Familiarity and Prompt Specificity on Citation Fabrication in Mental Health Research Using Large Language Models: Experimental Study (2025) · cites this paper as [4]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“However, while most LLM adopters reported that these tools enhanced research efficiency and improved the quality of their work, many also expressed concerns about inaccuracies, misleading content, and biases in the responses generated by these models [4].”
From Influence of Topic Familiarity and Prompt Specificity on Citation Fabrication in Mental Health Research Using Large Language Models: Experimental Study (2025) · cites this paper as [4]
The cited source says what the claim attributes to it.
In the citing paper
“A recent study showed that nearly 70% of mental health scientists reported using ChatGPT to assist with research-related tasks, including writing and drafting, coding, and administrative support [4].”
From Influence of Topic Familiarity and Prompt Specificity on Citation Fabrication in Mental Health Research Using Large Language Models: Experimental Study (2025) · cites this paper as [4]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Abstract > Findings
“Common challenges reported by users included inaccurate responses (78%), ethical concerns (48%) and biased outputs (27%).”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Findings > User perspectives
“ethical concerns (n=227; 45.8%)”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“X’s Grok (n=9; 1.3%)”
§ Role of large language models in mental health research: an international survey of researchers’ practices and perspectives > Findings > User perspectives
Reported 1.3 · recomputed 1.814516
Citations
Cites
ChatGPT for good? On opportunities and challenges of large language models for education
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., et al. (2023) · Learning and Individual Differences · doi:10.1016/j.lindif.2023.102274 · Cited by 7,326
Large language models in medicine
Thirunavukarasu, A. J., Ting, D. S. J., Elangovan, K., Gutierrez, L., Tan, T. F., & Ting, D. S. W. (2023) · Nature Medicine · doi:10.1038/s41591-023-02448-8 · Cited by 3,944 · cited 1×
“3 LLMs have been widely studied in education, 4 medicine 5 and engineering research, 6 yet their empirical investigation within mental health settings has received comparatively less attention.”
Practical and ethical challenges of large language models in education: A systematic scoping review
Yan, L., Sha, L., Zhao, L., Li, Y., Martinez‐Maldonado, R., Chen, G., Li, X., Jin, Y., & Gašević, D. (2024) · British Journal of Educational Technology · doi:10.1111/bjet.13370 · Cited by 1,435
Large Language Models for Software Engineering: A Systematic Literature Review
Hou, X., Zhao, Y., Liu, Y., Yang, Z., Wang, K., Li, L., Luo, X., Lo, D., Grundy, J., & Wang, H. (2024) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3695988 · Cited by 1,029
Chelli, M., Descamps, J., Lavoué, V., Trojani, C., Azar, M., Deckert, M., Raynier, J.-L., Clowez, G., Boileau, P., & Ruetsch-Chelli, C. (2024) · Journal of Medical Internet Research · doi:10.2196/53164 · Cited by 602 · cited 1×
“20 These findings also raise concerns about whether less experienced researchers, who appear to use LLMs more frequently, are adequately equipped to detect and correct errors generated by these tools.”
AI and science: what 1,600 researchers think
Van Noorden, R., & Perkel, J. M. (2023) · Nature · doi:10.1038/d41586-023-02980-0 · Cited by 383
Torous, J., Linardon, J., Goldberg, S. B., Sun, S., Bell, I., Nicholas, J., Hassan, L., Hua, Y., Milton, A., & Firth, J. (2025) · World Psychiatry · doi:10.1002/wps.21299 · Cited by 323 · cited 1×
“7 By analysing vast amounts of data derived from clinical case notes, session transcripts,”
Self-citationRandomized Trial of a Generative AI Chatbot for Mental Health Treatment
Heinz, M. V., Mackin, D. M., Trudeau, B. M., Bhattacharya, S., Wang, Y., Banta, H. A., Jewett, A. D., Salzhauer, A. J., Griffin, T. Z., & Jacobson, N. C. (2025) · NEJM AI · doi:10.1056/aioa2400802 · Cited by 287
Large Language Models for Mental Health Applications: Systematic Review
Guo, Z., Lai, A., Thygesen, J. H., Farrington, J., Keen, T., & Li, K. (2024) · JMIR Mental Health · doi:10.2196/57400 · Cited by 269
The Opportunities and Risks of Large Language Models in Mental Health
Lawrence, H. R., Schneider, R. A., Rubin, S. B., Matarić, M. J., McDuff, D. J., & Jones Bell, M. (2024) · JMIR Mental Health · doi:10.2196/59479 · Cited by 235 · cited 1×
“22 Furthermore, inequitable access to advanced LLM platforms, due to cost, infrastructure or policy barriers, may widen global research disparities, limiting who can fully benefit from these tools.”
Research works printed but not linked: 1
- Naveed H , Wang Y , Malik M , et al . A comprehensive overview of large language models. arXiv 2023.
Cited by
Linardon, J., Jarman, H. K., McClure, Z., Anderson, C., Liu, C., & Messer, M. (2025) · JMIR Mental Health · doi:10.2196/80371 · Cited by 40
“We selected GPT-4o because it is the most recent and widely adopted LLM among academic researchers [4], making our results relevant to a significant proportion of users.”
InfluentialSelf-citationVoultsiou, E., & Moussiades, L. (2026) · Electronics · doi:10.3390/electronics15030524 · Cited by 27
Gong, E. J., Bang, C. S., & Shin, Y. S. (2026) · Bioengineering · doi:10.3390/bioengineering13030365 · Cited by 17
“Recent surveys of mental health researchers found that 69.5% have used LLMs, though fewer than 15% employ them for complex analytical tasks such as data analysis or study design [11].”
InfluentialLinardon, J., Thomas, J. J., Crow, S. J., Ghaderi, A., Hilbert, A., Klump, K. L., Wade, T. D., Walsh, B. T., & Weissman, R. S. (2025) · International Journal of Eating Disorders · doi:10.1002/eat.24543 · Cited by 8
Self-citationThe AI interviewer: multi-faceted evaluation of adaptive questioning by large language models
Panfilova, A. S., Bolshev, V., Mozikov, M., Латынов, В. В., Vanin, A. V., Nestik, T. A., Nourkova, V. V., Александровна, В. А., Kozin, M., Serohvostov, A., et al. (2026) · Scientific Reports · doi:10.1038/s41598-026-46517-7 · Cited by 5
Metadata
- Authors
- Linardon, J., Messer, M., Anderson, C., Liu, C., McClure, Z., Jarman, H. K., Goldberg, S. B., & Torous, J.
- Affiliations
- Deakin University (AU)
- University of Wisconsin–Madison (US)
- Beth Israel Deaconess Medical Center (US)
- Corresponding author
- Linardon, Jake
- ORCID
- Venue
- BMJ Mental Health
- Publisher
- BMJ
- Published
- 2025-06
- Volume
- 28
- Issue
- 1
- Pages
- e301787
- ISSN
- 2755-9734 (ISSN-L)
- PMC
- PMC12164621
- OpenAlex
- W4411295240
- Cited by
- 10
- References
- 24
- Citations by year
- 2026: 8 · 2025: 2
- FWCI
- 2.39
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Locations
- BMJ Mental Health (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Figshare (submittedVersion, cc-by)
- Funders
- National Health and Medical Research Council (APP1196948)
- NCCIH (K23AT010879, R24AT012845)
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- large language models, mental health research, ChatGPT, researcher attitudes, research ethics, AI-generated text, code generation, early-career researchers, responsible AI use, research evaluation
- MeSH terms
- Large Language Models, Female, Humans, Language, Male, Mental Health, Surveys and Questionnaires, Research Personnel, Biomedical Research
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Role of large language models in mental health research: an international survey of researchers' practices and perspectives. https://app.cochat.ai/papers/10.1136/bmjment-2025-301787
- The paper
- Linardon, J., Messer, M., Anderson, C., Liu, C., McClure, Z., Jarman, H. K., Goldberg, S. B., & Torous, J. (2025). Role of large language models in mental health research: an international survey of researchers' practices and perspectives. BMJ Mental Health. https://doi.org/10.1136/bmjment-2025-301787
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.
