Pith. sign in

REVIEW 4 major objections 7 minor 42 references

Advancing Conversational Psychotherapy: Integrating Privacy, Dual-Memory, and Domain Expertise with Large Language Models

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a privacy-preserving chatbot with dual memory and therapist-response retrieval can generate responses a trained preference model ranks comparable to licensed therapists, though below the best human answers.

desk verdict SoulSpeak is a credible integration of known pieces with candid limitations, but the 'comparable to licensed therapists' claim rests on worst/best comparisons and an unvalidated preference model, so the central comparison is underdetermined. read the letter →

arxiv 2412.02987 v1 pith:673P44UA submitted 2024-12-04 cs.CL cs.CY

XiuYu Zhang , Zening Luo This is my paper · ORCID
classification cs.CLcs.CY
keywords conversationalpsychotherapylargelanguagemodelsretrieval-augmentedgenerationlong-termmemoryprivacyanonymizationpreferencemodelingmentalhealthchatbot
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that a chat agent can approach the level of human therapists in text-based psychotherapy if it combines three things: stripping personal identifiers before storage or API calls, remembering people and events across sessions through a long-term entity store, and retrieving example therapist answers to inform generation. On 113 questions, responses from the system built on GPT-3.5-turbo are preferred by a trained preference model over the lowest-rated human therapist answers and are comparable to mid-ranked ones, though human therapists remain preferred at the top. The long-term memory module raises relevance to GPT-4-generated reference answers from about 0.41 to 0.65 compared with a no-memory baseline. The paper also introduces CPPM, a reusable preference model for evaluating psychotherapy-focused language models from therapist-response upvote and view data. If the preference model's simulation is right, this is evidence that low-cost, private, memory-capable chatbots could offer a viable middle ground in mental-health access.

What carries the argument

The argument rides on two mechanisms. First, CPPM: a BERT binary classifier that takes two responses to the same question and predicts which one a service user would prefer, trained on pairwise labels built from the preference score $\log(\text{upvotes}+1)/\log(\text{views}+1)$ computed from therapist responses on a public counseling site. Second, the retrieval pipeline: an entity store that accumulates LLM-generated summaries of people and events across sessions, plus short-term conversation history, with retrieved therapist answers injected into the prompt only when the cosine similarity between the user query and a stored question exceeds the threshold $\alpha = 0.2$. The claim of therapist-level quality is produced by the former; the claimed continuity and personalization are produced by the latter.

What would settle it

Conduct a human preference study in which people seeking mental-health support compare anonymized pairs of therapist and SOUL SPEAK responses to the same questions; if their choices match CPPM no better than chance, or if human raters prefer the therapist's answer far more often than CPPM predicts, the claim that SOUL SPEAK produces therapist-comparable responses loses its empirical support.

Watch

Extended reading notes

Core claim

The central discovery is that a psychotherapy chatbot assembled from an LLM, a privacy anonymization layer, a dual-memory system, and retrieval from a corpus of therapist-client exchanges can generate single-turn responses that, under the paper's preference simulation, are chosen over the lowest-scoring therapist responses for the same questions and are comparable to mid-ranked ones. The paper states this as evidence that SOUL SPEAK is able to produce responses comparable to licensed professional therapists. The strongest prompt variant, gkpPsychoTherapy, is preferred over the human-worst answers in 82.30% of comparisons, while all SOUL SPEAK prompts stay below the human-best answers, which CPPM prefers in 2.65% to 15.93% of comparisons. Long-term memory further lifts relevance: with memory enabled, responses score 0.6495 against GPT-4 reference answers versus 0.4089 without it. The authors also show that a fine-tuned BERT preference model, CPPM, can serve as a reusable simulator of service-user preference for evaluating psychotherapy language models.

Load-bearing premise

All the headline comparisons to licensed therapists pass through the simulated preference model CPPM, which assumes that the ratio of logged upvotes to logged views on one counseling website stands in for what real service users prefer, including for text written by a chatbot.

Editorial extensions

If this is right

  • A deployed chatbot using these modules would keep personal names, locations, and dates out of stored memory and third-party API calls, storing only anonymized entity summaries.
  • Long-term memory allows the chatbot to reference earlier sessions, such as a co-worker's past behavior, instead of asking the user to re-explain, which is the kind of continuity therapists provide.
  • CPPM can be repurposed as a low-cost training and evaluation signal for other mental-health language models, reducing dependence on expensive human preference data.
  • Under the preference simulation, prompt design shifts outcomes less than the presence of retrieval and memory; all evaluated prompt variants beat the lowest-rated therapist answers.
  • Stronger instruction-tuned models are not automatically better: GPT-4o's responses were less preferred than GPT-3.5's, often refusing to give suggestions or listing points, so psychotherapy prompting may need model-specific tuning.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper's single-turn evaluation, a natural extension would be to test whether anonymization degrades the very contextual cues the entity store relies on, by comparing responses with and without the privacy layer.
  • The memory evaluation uses GPT-4-generated scenarios and summaries as ground truth; a longitudinal study with real service users would show whether entity recall improves perceived empathy or clinical outcomes, not just cosine similarity.
  • Because CPPM's labels come from upvote and view popularity on one counseling site, using it as a reward model would inherit that site's popularity bias, potentially favoring generic advice over personalized nuance.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The paper presents SOUL SPEAK, an LLM-based conversational psychotherapy system that combines a privacy module (NER-based anonymization of personally identifiable information), a dual-memory module (a short-term conversation window plus a long-term entity store), and a retrieval-augmented generation pipeline over the Counsel Chat dataset of therapist-client exchanges. The authors introduce two fine-tuned BERT models: CPPM, which simulates service-user preference between two responses using preference scores derived from upvote/view counts, and a topic-classification model used to compute response relevance. The system is evaluated by comparing its responses with human therapist responses on 113 questions from Counsel Chat, by testing the long-term memory module on 16 GPT-4-generated scenarios, and by examining the privacy module qualitatively. The central claim is that SOUL SPEAK produces responses comparable to those of licensed professional therapists.

Significance. If the evaluation were fully supported, the paper would make a useful contribution to applied LLM-based mental-health support: the modular architecture is clearly described, the inclusion of privacy anonymization and long-term memory addresses two recognized weaknesses of therapy chatbots, and the CPPM could be a reusable evaluation tool. The authors provide model-training details, prompt templates, and appendix material that make the system reasonably reproducible. The paper also honestly acknowledges the absence of human evaluation and the preliminary nature of the study. However, the significance of the headline claim is currently limited by the self-referential nature of the evaluation: CPPM is trained on the same dataset that supplies the knowledge base and the human comparison set, and the long-term memory test uses synthetically generated scenarios and LLM-generated ground truth. The contribution is better framed as a system design with a preliminary, simulation-based evaluation rather than as a demonstrated equivalence with human therapists.

major comments (4)
  1. [Section 4.1 and Appendix E] The claim that SOUL SPEAK produces responses 'comparable to licensed professional therapists' is not supported by the worst/best comparison design. Beating the lowest-scored therapist response and losing to the highest-scored response is compatible with performing far below the median therapist. Appendix E reports 'human-medium' win rates for GPT-3.5 ranging from 41.59% to 59.29% across prompts, which is around chance and sometimes below chance, yet the main text omits this comparison. The paper should report the win rate against a representative or median human response, along with confidence intervals, and should temper the headline if that win rate is near chance.
  2. [Section 4.1, Appendix A] The validity of CPPM is load-bearing and is assumed rather than demonstrated. Preference labels are computed as log(upvotes+1)/log(views+1) on counselchat.com, where upvotes range from 0 to 12, and the set of upvotes is extremely sparse. CPPM is trained on the same Counsel Chat data that also provides the knowledge-base retrievals and the human comparison set, and its 97% validation accuracy is measured on a split of this same preference-construction procedure, not against independent human preference judgments. The paper explicitly states 'assuming that the simulation by CPPM is representative of the true distribution of preferences.' A concrete test is needed: collect a small set of independent human preference annotations, especially for LLM-generated responses, and report CPPM's agreement with those annotations. Without such evidence, the comparison to licensed therapists remains a simulation-based plausibility check rather than an empirical result.
  3. [Section 5, Table 1] The long-term memory evaluation is based on only 16 GPT-4-generated scenarios, with no error bars, no per-scenario distribution, and no statistical significance test for the improvement in average relevance from 0.4089 to 0.6495. Moreover, the 'ground truth' sample answers and key-information summaries are also generated by GPT-4, so the improved cosine similarity to those targets may reflect stylistic closeness to GPT-4 rather than clinically meaningful memory retrieval. The paper should report the per-scenario results, a paired test, and an evaluation against independently written or human-validated ground truth. This issue directly affects the claim that the long-term memory module 'effectively enriches' the agent's responses.
  4. [Section 6] The limitations section acknowledges that 'human evaluation would be valuable, but due to the special conditions and privacy concerns associated with psychotherapy, we were unable to conduct human evaluations on a meaningful scale' and that the study 'does not necessarily reflect its long-term efficacy.' These statements are in tension with the earlier claims, in the same section, that 'Experimental results prove the effectiveness of SOUL SPEAK as a possible alternative to traditional therapy sessions' and, in Section 4.1, that the preference evaluation demonstrates comparability with licensed therapists. The paper should align its conclusions with its own stated limitations, for example by describing the results as preliminary simulation-based evidence rather than proof of effectiveness.
minor comments (7)
  1. [Section 4.1] The text says 'as demonstrated in Figure 3' when describing the construction of CPPM training pairs; this should refer to Figure 2, which shows the CPPM construction pipeline.
  2. [Section 4.1] The phrase 'we decide to train a model to stimulate the human preference' should be 'simulate the human preference.'
  3. [Section 3.1] The text says 'name entity recognition'; the standard term is 'named entity recognition.'
  4. [Throughout] The system name is written as 'SOUL SPEAK' in most places but also as 'Soulspeak' and 'SoulSpeak'; the notation should be made consistent.
  5. [Section 3.3] The formatting 'n = 10by default' is missing a space.
  6. [Table 1 and Figure 3] No confidence intervals or measures of variance are reported for the preference percentages or the average relevance scores; for a quantitative comparison, these should be included.
  7. [Appendix E] The 'human-medium' row is given without a corresponding figure or statistical comparison in the main text; the authors should decide whether this is a primary result or a robustness check and present it accordingly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the evaluation rests on an acknowledged proxy preference model and proxy ground truth, not on a construction that reduces the conclusion to its inputs.

full rationale

The paper's central claim is that SoulSpeak produces responses comparable to licensed professional therapists, supported by CPPM, a fine-tuned BERT preference model trained on Counsel Chat preference scores. This is an empirical evaluation, not a derivation whose conclusion is equivalent to its inputs by construction. SoulSpeak's outputs are novel text not present in the CPPM training set, and CPPM is a learned function rather than a fitted parameter renamed as a prediction. The comparison to 'human-worst' and 'human-best' anchors is not guaranteed by the preference-score formula: Figure 3a reports preference rates of 69.91%-82.30% against human-worst, not 100%, so the outcome is not forced. The paper explicitly acknowledges the load-bearing assumption: 'assuming that the simulation by CPPM is representative of the true distribution of preferences' (Section 4.1), and the Limitations section states that human evaluation was not conducted and CPPM was used instead. These are validity threats and correctness risks, not circular reductions. Similarly, the long-term memory evaluation uses GPT-4-generated scenarios and expected responses as ground truth against GPT-3.5-turbo outputs; this is an automated proxy, not the same system evaluating itself by construction. The knowledge base and test questions both come from the Counsel Chat dataset, which creates a closed evaluation loop and possible leakage, but the paper does not define the target result in terms of the evaluation metric in a way that makes the conclusion tautological. No self-citations carry load-bearing weight. Therefore, under the quoted-reduction standard, no significant circularity is present.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central results rest on three hand-chosen hyperparameters and five domain assumptions. None are unusual for an applied NLP paper, but the preference-score axiom and the CPPM-generalization axiom carry most of the evaluation weight.

free parameters (3)
  • Retrieval threshold alpha = 0.2
    Chosen by the authors after preliminary runs from 0.1 to 0.5 on the evaluation setting (Section 3.4), so the threshold is tuned to the same test context used for later comparisons.
  • Short-term memory window n = 10
    Set by default; the number of recent interactions appended to the prompt. Not fitted, but a hand-chosen hyperparameter affecting response quality.
  • Number of retrieved therapist responses k = 1
    k=1 in experiments; affects how much domain context is injected into the prompt.
assumptions (5)
  • domain assumption Upvote/view ratio approximates service-user preference
    Section 4.1 defines preference score as log(upvotes+1)/log(views+1) and uses it to label all CPPM training data; no evidence that this correlates with actual patient preference.
  • domain assumption CPPM generalizes from therapist-response pairs to comparisons involving LLM-generated responses and to real human preference
    Section 4.1 uses CPPM to compare SoulSpeak outputs against therapist answers, but CPPM was trained only on therapist-response pairs and was never validated on human ratings of generated text.
  • domain assumption GPT-4-generated scenarios and reference answers are valid ground truth
    Section 5 and Appendix D: 16 scripted scenarios plus expected therapist responses and key information are generated by GPT-4 and used as the benchmark for the long-term memory module.
  • domain assumption Counsel Chat public Q&A responses are representative of psychotherapy
    Appendix A: data is from counselchat.com, where therapists answer public questions with upvoting; this is not actual private therapy dialogue, but the paper treats it as a therapy dataset.
  • domain assumption spaCy NER precision of 85.16% is sufficient for privacy
    Section 3.1 relies on en_core_web_lg for PII detection with benchmark precision 85.16%, but no false-negative analysis on therapy-style input is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Advancing Conversational Psychotherapy: Integrating Privacy, Dual-Memory, and Domain Expertise with Large Language Models." pith.science (2026). https://pith.science/paper/673P44UA

@misc{pith2026241202987,
  author       = {Pith},
  title        = {Pith review of: Advancing Conversational Psychotherapy: Integrating Privacy, Dual-Memory, and Domain Expertise with Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/673P44UA}},
  note         = {Machine review of arXiv:2412.02987}
}
read the original abstract

Mental health has increasingly become a global issue that reveals the limitations of traditional conversational psychotherapy, constrained by location, time, expense, and privacy concerns. In response to these challenges, we introduce SoulSpeak, a Large Language Model (LLM)-enabled chatbot designed to democratize access to psychotherapy. SoulSpeak improves upon the capabilities of standard LLM-enabled chatbots by incorporating a novel dual-memory component that combines short-term and long-term context via Retrieval Augmented Generation (RAG) to offer personalized responses while ensuring the preservation of user privacy and intimacy through a dedicated privacy module. In addition, it leverages a counseling chat dataset of therapist-client interactions and various prompting techniques to align the generated responses with psychotherapeutic methods. We introduce two fine-tuned BERT models to evaluate the system against existing LLMs and human therapists: the Conversational Psychotherapy Preference Model (CPPM) to simulate human preference among responses and another to assess response relevance to user input. CPPM is useful for training and evaluating psychotherapy-focused language models independent from SoulSpeak, helping with the constrained resources available for psychotherapy. Furthermore, the effectiveness of the dual-memory component and the robustness of the privacy module are also examined. Our findings highlight the potential and challenge of enhancing mental health care by offering an alternative that combines the expertise of traditional therapy with the advantages of LLMs, providing a promising way to address the accessibility and personalization gap in current mental health services.

Figures

Figures reproduced from arXiv: 2412.02987 by the authors.

Figure 1
Figure 1. The SOULSPEAK system architecture and user workflow. users (Thompson and Harper, 2011). In this pa￾per and the scope of chatbots, the term service user is used as language model-powered agents are not licensed mental practitioners who can give medical treatment despite the ongoing debate over the usage of patient versus service users (Priebe, 2021). Through conversational psychotherapy, mental health therapists buil… view at source ↗
Figure 2
Figure 2. Overview of the Conversational Psychotherapy Preference Model (CPPM) to simulate service users’ [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Simulation of user preference on generated responses and therapists’ responses. Human-worst/human-best [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Statistical comparison between the therapist’s and S [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 6
Figure 6. Figure 6: Effectiveness of long-term memory mod￾ule. Unlike the baseline GPT-3.5 model, SOULSPEAK adeptly records and retrieves relevant details about the entity "Derek" from the entity store as shown in Fig￾ure 5. This functionality enables the agent to effectively incorporate …
Figure 5
Figure 5. Figure 5: Example of one entity and its summary in SOULSPEAK’s entity store. The entity name is anonymized by the privacy module. The entity sum￾mary is generated by an LLM API call given the context of the previous therapy session. system both qualitatively and quantitatively, …
Figure 9
Figure 9. Figure 9: Distribution of views in the dataset. Upvotes: The upvotes of question-answer pairs range from 0 to 12. See [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: Distribution of upvotes in the dataset [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: Relation between the number of views and [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: Topics covered in the dataset [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]
Figure 13
Figure 13. Figure 13: Topics covered in the dataset by percentages. [PITH_FULL_IMAGE:figures/full_fig_p014_13.png]
Figure 14
Figure 14. Figure 14: Loss of CPPM. 100 200 300 400 500 600 700 Step 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 Accuracy Validation Accuracy Validation Accuracy [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: Validation accuracy of CPPM. Training parameters: The model training was configured with the following parameters, aiming to optimize performance while ensuring efficient re￾source utilization. The training process took about 30 minutes on an NVIDIA P100. Batch Size: …
Figure 16
Figure 16. Figure 16: Loss of topic classification model. 100 200 300 400 500 600 700 Step 0.50 0.55 0.60 0.65 0.70 0.75 Accuracy Validation Accuracy Validation Accuracy [PITH_FULL_IMAGE:figures/full_fig_p018_16.png]
Figure 17
Figure 17. Figure 17: Validation accuracy of topic classification [PITH_FULL_IMAGE:figures/full_fig_p018_17.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 30 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    OpenAI Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, Red Avila, Igor Babuschkin, Suchir Balaji, Valerie Balcom, Paul Baltescu, Haiming Bao, Mo Bavarian, Jeff Belgum, Irwan Bello, Jake Berdine, Gabriel Bernadett-Shapiro, Christopher Berner, Lenny...

    work page 2023

  4. [4]

    Nor Aishah Ahad and Sharipah Soaad Syed Yahaya. 2014. https://api.semanticscholar.org/CorpusID:119832403 Sensitivity analysis of welch's t-test

    work page 2014

  5. [5]

    Tim Althoff, Kevin Clark, and Jure Leskovec. 2016. Large-scale analysis of counseling conversations: An application of natural language processing to mental health. Transactions of the Association for Computational Linguistics, 4:463 -- 476

    work page 2016

  6. [6]

    Nicolas Bertagnolli. 2020. Counsel chat: Bootstrapping high-quality therapy data. https://github.com/nbertagnolli/counsel-chat

    work page 2020

  7. [7]

    Siyuan Chen, Mengyue Wu, Ke Zhu, Kunyao Lan, Zhiling Zhang, and Lyuchun Cui. 2023. Llm-empowered chatbots for psychiatrist and patient simulation: Application and evaluation. ArXiv, abs/2305.13614

    arXiv 2023

  8. [8]

    Munmun De Choudhury and Sushovan De. 2014. Mental health discourse on reddit: Self-disclosure, social support, and anonymity. Proceedings of the International AAAI Conference on Web and Social Media

    work page 2014

Show all 42 references
  1. [9]

    Kyungmi Chung, Hee Young Cho, and Jin Young Park. 2020. A chatbot for perinatal women’s and partners’ obstetric and mental health care: Development and usability evaluation study. JMIR Medical Informatics, 9

    2020

  2. [10]

    Kerstin Denecke, Sayan Vaaheesan, and Aaganya Arulnathan. 2020. A mental health chatbot for regulating emotions (sermo) - concept and usability test. IEEE Transactions on Emerging Topics in Computing, 9:1170--1182

    2020

  3. [11]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In North American Chapter of the Association for Computational Linguistics

    2019

  4. [12]

    Explosion.ai. 2023. spaCy English Model (en\_core\_web\_lg) Release 3.7.1 . https://github.com/explosion/spacy-models/releases/tag/en_core_web_lg-3.7.1. Accessed: 12/11/2023

    2023

  5. [13]

    Farr, James J

    James N. Farr, James J. Jenkins, and Donald G. Paterson. 1951. https://doi.org/10.1037/h0062427 Simplification of flesch reading ease formula. Journal of Applied Psychology, 35(5):333–337

    1951 doi

  6. [14]

    Rudolph Flesch. 1948. https://doi.org/10.1037/h0057532 A new readability yardstick. Journal of Applied Psychology, 32(3):221–233

    1948 doi

  7. [15]

    Zhijun Guo, Alvina Lai, Johan H Thygesen, Joseph Farrington, Thomas Keen, and Kezhi Li. 2024. Large language model for mental health: A systematic review

    2024

  8. [16]

    M D Romael Haque and Sabirat Rubya. 2023. https://doi.org/10.2196/44838 An overview of chatbot-based mobile mental health apps: Insights from app description and user reviews . JMIR Mhealth Uhealth, 11:e44838

    2023 doi

  9. [17]

    Tianyu He, Guanghui Fu, Yi Jing Yu, Fan Wang, Jianqiang Li, Qing Zhao, Changwei Song, Hongzhi Qi, Dan Luo, Huijing Zou, and Bing Xiang Yang. 2023. Towards a psychological generalist ai: A survey of current applications of large language models and future prospects. ArXiv, abs/...

    2023 arXiv

  10. [18]

    Yuhao He, Li Yang, Xiaokun Zhu, Bin Wu, Shuo Zhang, Chunlian Qian, and Tian-Ci Tian. 2022. Mental health chatbot for young adults with depressive symptoms during the covid-19 pandemic: Single-blind, three-arm randomized controlled trial. Journal of Medical Internet Research, 24

    2022

  11. [19]

    Cardinal, Angus Roberts, Robert J Stewart, and Sumithra Velupillai

    Julia Ive, Natalia Viani, Joyce Kam, Lucia Yin, Somain Verma, Stephen Puntis, Rudolf N. Cardinal, Angus Roberts, Robert J Stewart, and Sumithra Velupillai. 2020. Generation and evaluation of artificial mental health records for natural language processing. NPJ Digital Medicine, 3

    2020

  12. [20]

    Langchain.ai. 2023. Langchain: An open-source library for building llm-powered applications. https://github.com/langchain-ai/langchain. Accessed: 2023-12-12

    2023

  13. [21]

    Howard Levene. 1961. https://api.semanticscholar.org/CorpusID:117424234 Robust tests for equality of variances

    1961

  14. [22]

    u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt\

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K\" u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt\" a schel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. In Adva...

    2020

  15. [23]

    Mental Health America . 2023. The state of mental health in america. https://mhanational.org/issues/state-mental-health-america

    2023

  16. [24]

    Nadim Nachar. 2007. https://api.semanticscholar.org/CorpusID:59357756 The mann ‐ whitney u: A test for assessing whether two independent samples come from the same distribution

    2007

  17. [25]

    OpenAI. 2024. https://openai.com/index/memory-and-new-controls-for-chatgpt/ Memory and new controls for chatgpt . Accessed: 2024-06-12

    2024

  18. [26]

    Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke E

    Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke E. Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Francis Christiano, Jan...

    2022 arXiv

  19. [27]

    Stefan Priebe. 2021. https://doi.org/10.1192/bjb.2021.40 Patients in mental healthcare should be referred to as patients and not service users . BJPsych Bulletin, 45(6):1–2

    2021 doi

  20. [28]

    Improving language understanding by generative pre-training

    Alec Radford, Karthik Narasimhan, Tim Salimans, and Ilya Sutskever. Improving language understanding by generative pre-training

  21. [29]

    Shapiro and M

    Samuel S. Shapiro and M. B. Wilk. 1965. https://api.semanticscholar.org/CorpusID:124868013 An analysis of variance test for normality (complete samples) . Biometrika, 52:591--611

    1965

  22. [30]

    Julia Stoll, Jonas M \"u ller, and Manuel Trachsel. 2020. Ethical issues in online psychotherapy: A narrative review. Frontiers in Psychiatry, 10

    2020

  23. [31]

    Ching Y. Suen. 1979. https://doi.org/10.1109/TPAMI.1979.4766902 n-gram statistics for natural language understanding and text processing . IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-1(2):164--172

    1979

  24. [32]

    Swift, Jennifer L

    Joshua K. Swift, Jennifer L. Callahan, Mick Cooper, and Susannah R. Parkin. 2018. The impact of accommodating client preference in psychotherapy: A meta-analysis. Journal of clinical psychology, 74 11:1924--1937

    2018

  25. [33]

    The Lancet Global Health . 2020. https://doi.org/10.1016/S2214-109X(20)30432-0 Mental health matters . Elsevier - PMC COVID-19 Collection

    2020 doi

  26. [34]

    Thompson and David Harper

    Andrew R. Thompson and David Harper. 2011. https://doi.org/10.1002/9781119973249.ch1 Introduction , chapter 1. John Wiley and Sons, Ltd

    2011 doi

  27. [35]

    Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N

    Ashish Vaswani, Noam M. Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Neural Information Processing Systems

    2017

  28. [36]

    Jing Wei, Sungdong Kim, Hyunhoon Jung, and Young-Ho Kim. 2023. Leveraging large language models to power chatbots for collecting user self-reported data. ArXiv, abs/2301.05843

    2023 arXiv

  29. [37]

    World Health Organization . 2022 a . Covid-19 pandemic triggers 25\ https://tinyurl.com/3skcx5d5

    2022

  30. [38]

    World Health Organization . 2022 b . Mental health. https://www.who.int/news-room/fact-sheets/detail/mental-health-strengthening-our-response

    2022

  31. [39]

    World Health Organization . 2023 a . Depressive disorder (depression). https://www.who.int/en/news-room/fact-sheets/detail/depression

    2023

  32. [40]

    World Health Organization . 2023 b . Suicide. https://www.who.int/news-room/fact-sheets/detail/suicide

    2023

  33. [41]

    Dey, and Dakuo Wang

    Xuhai Xu, Bingsheng Yao, Yuanzhe Dong, Saadia Gabriel, Hong Yu, James Hendler, Marzyeh Ghassemi, Anind K. Dey, and Dakuo Wang. 2024. https://doi.org/10.1145/3643540 Mental-llm: Leveraging large language models for mental health prediction via online text data . Proceedings of ...

    2024 doi

  34. [42]

    Tianlin Zhang, Annika Marie Schoene, Shaoxiong Ji, and Sophia Ananiadou. 2022. Natural language processing applied to mental illness detection: a narrative review. NPJ Digital Medicine, 5

    2022

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.