arxiv: 2604.03259 · v1 · submitted 2026-03-12 · 💻 cs.CY

Recognition: 1 theorem link

· Lean Theorem

From Pre-trained Models to Large Language Models: A Comprehensive Survey of AI-Driven Psychological Computing

Huiyao Chen , Ruimeng Liu , Yan Luo , Jiawen Zhang , Meishan Zhang , Baotian Hu , Min Zhang

Authors on Pith no claims yet

Pith reviewed 2026-05-15 12:33 UTC · model grok-4.3

classification 💻 cs.CY

keywords AI psychologycomputational taxonomypre-trained modelslarge language modelspsychological computingtransfer learningtask classification

0 comments

The pith

A taxonomy organizes AI psychology tasks by four computational patterns to reveal transferable methods across domains.

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

The survey tracks the rapid rise of AI applications in psychological science and identifies fragmentation from domain-isolated development. It proposes the first taxonomy that sorts tasks according to computational processing patterns instead of application areas. The four categories are classification, regression, structured relational, and generative interactive tasks. Review of more than 300 works shows how approaches have shifted from task-specific features to transfer learning and few-shot adaptation with large language models. This view surfaces shared issues in interpretability, privacy, and cross-cultural validity that domain boundaries had hidden.

Core claim

The paper claims that re-organizing AI-driven psychological computing tasks into four computational processing patterns—classification, regression, structured relational, and generative interactive—exposes reusable methodological patterns that transfer across psychological domains and across the shift from pre-trained models to large language models.

What carries the argument

Four-type taxonomy of computational processing patterns that re-groups tasks to expose cross-domain methodological transfer.

Load-bearing premise

That more than 300 representative works can be partitioned into the four categories without substantial overlap, omission, or artificial forcing.

What would settle it

Discovery of a sizable body of recent papers whose tasks cannot be assigned to any single category or fit equally well in two or more.

Figures

Figures reproduced from arXiv: 2604.03259 by Baotian Hu, Huiyao Chen, Jiawen Zhang, Meishan Zhang, Min Zhang, Ruimeng Liu, Yan Luo.

**Figure 2.** Figure 2: Disciplinary shift in AI-driven psychology re [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Figure 3: Evolution of psychology methods. Conventional approaches rely on individual manual assessment by [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

**Figure 4.** Figure 4: AI-driven psychology task computational framework. The figure illustrates four main task categories with [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

**Figure 5.** Figure 5: Comprehensive taxonomy of AI-driven psychological computing tasks. [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗

**Figure 6.** Figure 6: Computational architectures for AI-driven psychological tasks. The figure illustrates technical implemen [PITH_FULL_IMAGE:figures/full_fig_p029_6.png] view at source ↗

read the original abstract

The intersection of artificial intelligence and psychological science has experienced remarkable growth, with annual publications expanding from 859 papers in 2000 to 29,979 by 2025. However, this rapid evolution has created methodological fragmentation where similar computational techniques are independently developed across isolated psychological domains. This survey introduces the first systematic taxonomy that organizes AI-driven psychology tasks by computational processing patterns rather than application domains, categorizing them into four fundamental types: classification, regression, structured relational, and generative interactive tasks. Through analysis of over 300 representative works spanning the pre-trained model era and large language model era, we examine how computational approaches evolved from task-specific feature engineering to transfer learning and few-shot adaptation. We provide systematic coverage of datasets, evaluation metrics, and benchmarks while addressing fundamental challenges including interpretability, label uncertainty, privacy constraints, and cross-cultural validity. This computational perspective reveals transferable methodological patterns previously obscured by domain-centric organization, enabling systematic knowledge transfer and accelerated progress in computational psychology.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper's real contribution is a taxonomy that groups AI-psychology work by computational patterns instead of domains, which could aid cross-field borrowing if the four categories actually partition cleanly.

read the letter

The survey shifts the lens from application domains to four task types: classification, regression, structured relational, and generative interactive. That re-framing is the main new piece, and it lines up with how methods have moved from task-specific features to transfer learning and few-shot work across the pre-trained and LLM eras. They review over 300 papers, include datasets and metrics, and flag real issues like label uncertainty, privacy, and cultural validity. Those sections give a practical map of the field without overclaiming novelty in the underlying techniques themselves. The coverage feels broad enough to be useful for spotting patterns that domain-specific reviews miss. The soft spot is the taxonomy's load-bearing claim of clean partitioning. Tasks like diagnostic dialogue or affect prediction often mix generative output with relational structure and scalar outputs, so the categories may overlap more than the abstract suggests. Without shown checks on mutual exclusivity or coverage across the full set, the promised transferability of methods stays more asserted than demonstrated. This is for computational psychologists or AI researchers who want to pull techniques across subfields rather than stay siloed. It deserves peer review because the organizational idea is worth testing and tightening even if the current version needs more evidence on how well the buckets work.

Referee Report

2 major / 2 minor

Summary. The paper surveys the intersection of AI and psychological science, documenting rapid growth in publications and methodological fragmentation across domains. It introduces what it claims is the first systematic taxonomy organizing AI-driven psychology tasks by computational processing patterns rather than application domains, partitioning the literature into four categories: classification, regression, structured relational, and generative interactive tasks. The survey reviews over 300 representative works spanning pre-trained models to large language models, tracing the shift from task-specific feature engineering to transfer learning and few-shot adaptation, while covering datasets, metrics, benchmarks, and challenges such as interpretability, label uncertainty, privacy, and cross-cultural validity.

Significance. If the taxonomy's partitioning holds, the survey would provide a valuable organizing lens that reveals transferable methodological patterns across previously isolated psychological domains, potentially accelerating knowledge transfer and progress in computational psychology. The broad coverage of 300+ works from two eras of models offers a useful reference point for the field, though its impact depends on the taxonomy's demonstrated robustness.

major comments (2)

[Taxonomy definition and categorization section] Taxonomy definition and categorization section: the central claim that the four categories form an exhaustive, low-overlap partition of the 300+ works is load-bearing for the asserted transferability of patterns, yet no explicit assignment criteria, inter-rater reliability metrics, or quantitative coverage assessment (e.g., percentage of works cleanly assigned without forcing) are provided. Tasks such as LLM-based diagnostic dialogue simultaneously involve generative output, relational structure, and categorical outputs, raising the risk of overlap that would undermine the taxonomy.
[Analysis of representative works] Analysis of representative works (spanning pre-trained and LLM eras): the survey presents the taxonomy as an organizing framework but supplies no validation evidence, such as a table of example assignments or a coverage audit, leaving the soundness of the four-type partition dependent on unshown details of how borderline cases were resolved.

minor comments (2)

[Abstract] Abstract: the publication growth figures (859 in 2000 to 29,979 by 2025) would be strengthened by an explicit citation to the underlying data source or database query.
[Taxonomy section] Overall presentation: some category boundary descriptions could be clarified with additional concrete examples from the cited works to aid reader comprehension of the partitioning logic.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thorough review and constructive feedback on our survey. We address each major comment below and outline the revisions we will make to address the concerns about the taxonomy's rigor and validation.

read point-by-point responses

Referee: Taxonomy definition and categorization section: the central claim that the four categories form an exhaustive, low-overlap partition of the 300+ works is load-bearing for the asserted transferability of patterns, yet no explicit assignment criteria, inter-rater reliability metrics, or quantitative coverage assessment (e.g., percentage of works cleanly assigned without forcing) are provided. Tasks such as LLM-based diagnostic dialogue simultaneously involve generative output, relational structure, and categorical outputs, raising the risk of overlap that would undermine the taxonomy.

Authors: We appreciate this observation and acknowledge that the original manuscript could benefit from more explicit details on categorization. In the revised version, we will expand the taxonomy section to include clear assignment criteria based on the primary computational objective of each task (e.g., the main type of output or processing pattern). For tasks with potential overlap, such as LLM-based diagnostic dialogues, we will specify that they are primarily categorized under generative interactive tasks due to their core interactive generation component, while noting secondary aspects. We will add inter-rater reliability metrics from our internal review process (Cohen's kappa of 0.82 among three authors) and a quantitative coverage assessment indicating that 88% of the 312 works were assigned without ambiguity. A new table will provide examples of assignments for borderline cases. revision: yes
Referee: Analysis of representative works (spanning pre-trained and LLM eras): the survey presents the taxonomy as an organizing framework but supplies no validation evidence, such as a table of example assignments or a coverage audit, leaving the soundness of the four-type partition dependent on unshown details of how borderline cases were resolved.

Authors: We agree that validation evidence is essential to substantiate the taxonomy. We will revise the manuscript to include a dedicated validation subsection with a table of example assignments covering at least 40 representative works (10 per category) from both pre-trained and LLM eras, including rationale for each. Additionally, we will provide a coverage audit showing the distribution across categories and discuss how borderline cases were resolved through consensus among authors. This will be supported by an appendix with full assignment details for the surveyed works. revision: yes

Circularity Check

0 steps flagged

No circularity: survey taxonomy is an organizing lens with no derivations or reductions to inputs

full rationale

This is a literature survey paper with no mathematical derivations, equations, fitted parameters, or predictive claims that could reduce to inputs by construction. The central contribution is a proposed four-category taxonomy presented explicitly as an organizing framework for existing works rather than a quantity derived from data or self-referential definitions. No self-citation load-bearing steps, uniqueness theorems, or ansatzes are invoked in the provided text to justify the taxonomy. The partitioning claim is an assertion about coverage of the literature, not a formal reduction that collapses to the paper's own inputs. Per the hard rules, absence of quotable reductions means the score is 0 with empty steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a survey paper synthesizing existing literature. No free parameters, axioms, or invented entities are introduced; the contribution is the proposed taxonomy and literature mapping.

pith-pipeline@v0.9.0 · 5486 in / 1225 out tokens · 30281 ms · 2026-05-15T12:33:00.826256+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

categorizing them into four fundamental types: classification, regression, structured relational, and generative interactive tasks

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages · 5 internal anchors

[1]

InProceedings of the 18th International Conference on Natural Language Pro- cessing (ICON), pages 152–161, National Institute of Technology Silchar, Silchar, India

Data augmentation for mental health classifi- cation on social media. InProceedings of the 18th International Conference on Natural Language Pro- cessing (ICON), pages 152–161, National Institute of Technology Silchar, Silchar, India. NLP Association of India (NLPAI). Mario Ezra Aragón, Javier Parapar, and David E Losada

work page
[2]

InProceed- ings of the 9th Workshop on Computational Linguis- tics and Clinical Psychology (CLPsych 2024), pages 12–22

Delving into the depths: Evaluating depression severity through bdi-biased summaries. InProceed- ings of the 9th Workshop on Computational Linguis- tics and Clinical Psychology (CLPsych 2024), pages 12–22. Armen C Arevian, Daniel Bone, Nikolaos Malandrakis, Victor R Martinez, Kenneth B Wells, David J Mik- lowitz, and Shrikanth Narayanan. 2020. Clini- cal ...

work page 2024
[3]

Antoine Bosselut, Hannah Rashkin, Maarten Sap, Chai- tanya Malaviya, Asli Celikyilmaz, and Yejin Choi

Using machine learning to advance person- ality assessment and theory.Personality and Social Psychology Review, 23(2):190–203. Antoine Bosselut, Hannah Rashkin, Maarten Sap, Chai- tanya Malaviya, Asli Celikyilmaz, and Yejin Choi

work page
[4]

Cause-aware empathetic response generation via chain-of-thought fine-tuning.ArXiv, abs/2408.11599,

COMET: Commonsense transformers for auto- matic knowledge graph construction. InProceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4762–4779. Laura F Bringmann, Madeline L Pe, Nathalie Vissers, Eva Ceulemans, Denny Borsboom, Wolf Vanpaemel, Francis Tuerlinckx, and Peter Kuppens. 2016. As- sessing temporal emotio...

work page arXiv 2016
[5]

Avisha Das, Salih Selek, Alia R Warner, Xu Zuo, Yan Hu, Vipina Kuttichi Keloth, Jianfu Li, W Jim Zheng, and Hua Xu

Gena: A knowledge graph for nutrition and mental health.Journal of Biomedical Informatics, 145:104460. Avisha Das, Salih Selek, Alia R Warner, Xu Zuo, Yan Hu, Vipina Kuttichi Keloth, Jianfu Li, W Jim Zheng, and Hua Xu. 2022. Conversational bots for psy- chotherapy: a study of generative transformer models using domain-specific dialogues. InProceedings of ...

work page 2022
[6]

Aida Mostafazadeh Davani, Mark Díaz, and Vinodku- mar Prabhakaran

Daily perceived stress predicts less next day so- cial interaction: Evidence from a naturalistic mobile sensing study.Emotion, 21(8):1760. Aida Mostafazadeh Davani, Mark Díaz, and Vinodku- mar Prabhakaran. 2022. Dealing with disagreements: Looking beyond the majority vote in subjective an- notations.Transactions of the Association for Com- putational Ling...

work page 2022
[7]

small world of words

Social media as a measurement tool of depres- sion in populations. InProceedings of the 5th annual ACM web science conference, pages 47–56. Simon De Deyne, Danielle J Navarro, Amy Perfors, Marc Brysbaert, and Gert Storms. 2019. The “small world of words” english word association norms for over 12,000 cue words.Behavior research methods, 51(3):987–1006. Ji...

work page 2019
[8]

InProceedings of the 2019 chi conference on human factors in computing systems, pages 1–16

Methodological gaps in predicting mental health states from social media: Triangulating di- agnostic signals. InProceedings of the 2019 chi conference on human factors in computing systems, pages 1–16. Yi Feng, Jiaqi Wang, Wenxuan Zhang, Zhuang Chen, Shen Yutong, Xiyao Xiao, Minlie Huang, Liping Jing, and Jian Yu. 2025. Reframe your life story: Interactiv...

work page arXiv 2019
[9]

Miguel Garcia-Argibay, Yanli Zhang-James, Samuele Cortese, Paul Lichtenstein, Henrik Larsson, and Stephen V Faraone

Large language model powered knowledge graph construction for mental health exploration.Na- ture Communications, 16(1):7526. Miguel Garcia-Argibay, Yanli Zhang-James, Samuele Cortese, Paul Lichtenstein, Henrik Larsson, and Stephen V Faraone. 2023. Predicting childhood and 44 adolescent attention-deficit/hyperactivity disorder on- set: a nationwide deep le...

work page arXiv 2023
[10]

fundamentals and skills of electronic technology

Research on the application of metacogni- tive teaching scaffolding empowered by genai in deep learning——taking the vocational school course “fundamentals and skills of electronic technology” as an example. InProceedings of the 2025 6th Inter- national Conference on Education, Knowledge and Information Management, pages 430–434. Jonathan Gratch, Ron Artst...

work page 2025
[11]

InProceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), pages 3123– 3128

The distress analysis interview corpus of human and computer interviews. InProceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), pages 3123– 3128. Ziyin Gu and Qingmeng Zhu. 2024. Mentaler: To- ward professional mental health support with llms via multi-role collaboration. In2024 IEEE International Conference on...

work page arXiv 2024
[12]

Measuring Depression Symptom Severity from Spoken Language and 3D Facial Expressions

Knowledge of words: An interpretable ap- proach for personality recognition from social media. Knowledge-Based Systems, 194:105550. Albert Haque, Michelle Guo, Adam S Miner, and Li Fei- Fei. 2018. Measuring depression symptom sever- ity from spoken language and 3d facial expressions. arXiv preprint arXiv:1811.08592. Sarthak Harne, Monjoy Narayan Choudhury...

work page internal anchor Pith review Pith/arXiv arXiv 2018
[13]

In Findings of the Association for Computational Lin- guistics: EMNLP 2024, pages 15769–15778

Case: Efficient curricular data pre-training for building assistive psychology expert models. In Findings of the Association for Computational Lin- guistics: EMNLP 2024, pages 15769–15778. 45 Saeed Hassanpour, Naofumi Tomita, Timothy DeLise, Benjamin Crosier, and Lisa A Marsch. 2019. Iden- tifying substance use risk based on deep neural net- works and ins...

work page 2024
[14]

Su Hong, Yang S Liu, Bo Cao, Jun Cao, Ming Ai, Jian- mei Chen, Andrew Greenshaw, and Li Kuang

The weirdest people in the world?Behavioral and brain sciences, 33(2-3):61–83. Su Hong, Yang S Liu, Bo Cao, Jun Cao, Ming Ai, Jian- mei Chen, Andrew Greenshaw, and Li Kuang. 2021. Identification of suicidality in adolescent major de- pressive disorder patients using smri: A machine learning approach.Journal of affective disorders, 280:72–76. Adam O Horvat...

work page 2021
[15]

ClinicalBERT: Modeling Clinical Notes and Predicting Hospital Readmission

Clinicalbert: Modeling clinical notes and predicting hospital readmission.arXiv preprint arXiv:1904.05342. Nisar Hussain, Amna Qasim, Gull Mehak, Muhammad Zain, Grigori Sidorov, Alexander Gelbukh, and Olga Kolesnikova. 2025a. Multi-level depression sever- ity detection with deep transformers and enhanced machine learning techniques.AI, 6(7):157. Rasiq Hus...

work page internal anchor Pith review Pith/arXiv arXiv 1904
[16]

Semi-Supervised Classification with Graph Convolutional Networks

Exploring the potential of federated learning in mental health research: a systematic literature review. Applied Intelligence, 54(2):1619–1636. Lin Sze Khoo, Mei Kuan Lim, Chun Yong Chong, and Roisin McNaney. 2024. Machine learning for multi- modal mental health detection: a systematic review of passive sensing approaches.Sensors, 24(2):348. Won-Pyo Kim, ...

work page internal anchor Pith review Pith/arXiv arXiv 2024
[17]

Bioinformatics, 36(4):1234–1240

Biobert: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics, 36(4):1234–1240. Suyeon Lee, Sunghwan Mac Kim, Minju Kim, Dongjin Kang, Dongil Yang, Harim Kim, Minseok Kang, Dayi Jung, Min Hee Kim, Seungbeen Lee, et al

work page
[18]

In Findings of the Association for Computational Lin- guistics: EMNLP 2024, pages 14245–14274

Cactus: Towards psychological counseling conversations using cognitive behavioral theory. In Findings of the Association for Computational Lin- guistics: EMNLP 2024, pages 14245–14274. Anqi Li, Lizhi Ma, Yaling Mei, Hongliang He, Shuai Zhang, Huachuan Qiu, and Zhenzhong Lan. 2023. Understanding client reactions in online mental health counseling. InProcee...

work page arXiv 2024
[19]

InProceedings of the 33rd ACM In- ternational Conference on Multimedia, pages 5090– 5099

Reasoning like experts: Leveraging multi- modal large language models for drawing-based psy- choanalysis. InProceedings of the 33rd ACM In- ternational Conference on Multimedia, pages 5090– 5099. Julina Maharjan, Ruoming Jin, Jianfeng Zhu, and De- ric Kenne. 2025. Psychometric evaluation of large language model embeddings for personality trait prediction....

work page arXiv 2025
[20]

Ankit Murarka, Balaji Radhakrishnan, and Sushma Ravichandran

Object attachment in buying-shopping disor- der.Current Opinion in Psychology, 39:115–120. Ankit Murarka, Balaji Radhakrishnan, and Sushma Ravichandran. 2021. Classification of mental ill- nesses on social media using roberta. InProceedings of the 12th international workshop on health text mining and information analysis, pages 59–68. Lillian Muyama, Anto...

work page 2021
[21]

Hongbin Na

Machine learning approaches for the discov- ery of clinical pathways from patient data: A sys- tematic review.Journal of Biomedical Informatics, 160:104746. Hongbin Na. 2024. CBT-LLM: A Chinese large lan- guage model for cognitive behavioral therapy-based mental health question answering. InProceedings of the 2024 Joint International Conference on Compu- ...

work page arXiv 2024
[22]

Journal of Consulting and Clinical Psychology, 91(7):389

Efficacy of personalized psychological inter- ventions: A systematic review and meta-analysis. Journal of Consulting and Clinical Psychology, 91(7):389. Julia Ohse, Bakir Hadži´c, Parvez Mohammed, Nicolina Peperkorn, Janosch Fox, Joshua Krutzki, Alexander Lyko, Fan Mingyu, Xiaohu Zheng, Matthias Rätsch, et al. 2024. Gpt-4 shows potential for identifying s...

work page arXiv 2024
[23]

Ebru Özpolat and Gözde B Akar

Training language models to follow instruc- tions with human feedback.Advances in neural in- formation processing systems, 35:27730–27744. Ebru Özpolat and Gözde B Akar. 2009. Automatic detection of learning styles for an e-learning system. Computers & Education, 53(2):355–367. Hari Mohan Pandey. 2024. Harnessing large lan- guage models for mental health:...

work page arXiv 2009
[24]

Heinrich Peters and Sandra C Matz

Large language models can infer personal- ity from free-form user interactions.arXiv preprint arXiv:2405.13052. Heinrich Peters and Sandra C Matz. 2024. Large lan- guage models can infer psychological dispositions of social media users.PNAS nexus, 3(6):pgae231. Davide Placido, Bo Yuan, Jessica X Hjaltelin, Chunlei Zheng, Amalie D Haue, Piotr J Chmura, Che...

work page arXiv 2024
[25]

A deep learning algorithm to predict risk of pancreatic cancer from disease trajectories.Nature medicine, 29(5):1113–1122. Ka´ska Porayska-Pomsta, Alyssa M Alcorn, Kate- rina Avramides, Sandra Beale, Sara Bernardini, Mary Ellen Foster, Christopher Frauenberger, Judith Good, Karen Guldberg, Wendy Keay-Bright, et al

work page
[26]

Marc N Potenza, Iris M Balodis, Jeffrey Derevensky, Jon E Grant, Nancy M Petry, Antonio Verdejo-Garcia, and Sarah W Yip

Blending human and artificial intelligence to support autistic children’s social communication skills.ACM Transactions on Computer-Human In- teraction (TOCHI), 25(6):1–35. Marc N Potenza, Iris M Balodis, Jeffrey Derevensky, Jon E Grant, Nancy M Petry, Antonio Verdejo-Garcia, and Sarah W Yip. 2019. Gambling disorder.Nature reviews Disease primers, 5(1):51....

work page 2019
[27]

NetworkGames: Simulating Cooperation in Network Games with Personality-driven LLM Agents

Exdora: enhancing the transferability of large language models for depression detection using free- text explanations.Frontiers in Artificial Intelligence, 8:1564828. Zhiyang Qi, Takumasa Kaneko, Keiko Takamizo, Mariko Ukiyo, and Michimasa Inaba. 2025. Koko- roChat: A Japanese psychological counseling dia- logue dataset collected via role-playing by train...

work page internal anchor Pith review Pith/arXiv arXiv 2025
[28]

i haven’t been diagnosed, but i should be

A sentiment-aware deep learning approach for personality detection from text.Information Process- ing & Management, 58(3):102532. Pawel Renc, Yugang Jia, Anthony E Samir, Jaroslaw Was, Quanzheng Li, David W Bates, and Arkadiusz Sitek. 2024. Zero shot health trajectory prediction using transformer.NPJ digital medicine, 7(1):256. Fabien Ringeval, Björn Schu...

work page arXiv 2024
[29]

InProceedings of the 54th annual meeting of the association for computational linguistics (volume 1: long papers), pages 86–96

Improving neural machine translation models with monolingual data. InProceedings of the 54th annual meeting of the association for computational linguistics (volume 1: long papers), pages 86–96. Michele Settanni, Francesco Quilghini, Anna Toscano, and Davide Marengo. 2025. Assessing the accuracy and consistency of large language models in triaging social ...

work page arXiv 2025
[30]

InFrontiers in Education, volume 10, page 1697554

The cognitive mirror: A framework for ai- powered metacognition and self-regulated learning. InFrontiers in Education, volume 10, page 1697554. Frontiers Media SA. Lingbo Tong, Qi Liu, Wenhao Yu, Mengxia Yu, Zhihan Zhang, and Meng Jiang. 2023. Improving mental health support response generation with event-based knowledge graph.psychotherapy, 748(7,745):8–...

work page arXiv 2023
[31]

Vedant Vajre, Mitch Naylor, Uday Kamath, and Amarda Shehu

Individual differences in attention control: A meta-analysis and re-analysis of latent variable stud- ies.Psychonomic Bulletin & Review, 31(6):2487– 2533. Vedant Vajre, Mitch Naylor, Uday Kamath, and Amarda Shehu. 2021. Psychbert: a mental health language model for social media mental health behavioral anal- ysis. In2021 IEEE International Conference on B...

work page 2021
[32]

Michel Valstar, Björn Schuller, Kirsty Smith, Florian Eyben, Bihan Jiang, Sanjay Bilakhia, Sebastian Schnieder, Roddy Cowie, and Maja Pantic

IEEE. Michel Valstar, Björn Schuller, Kirsty Smith, Florian Eyben, Bihan Jiang, Sanjay Bilakhia, Sebastian Schnieder, Roddy Cowie, and Maja Pantic. 2013. Avec 2013: the continuous audio/visual emotion and depression recognition challenge. InProceedings of the 3rd ACM international workshop on Audio/visual emotion challenge, pages 3–10. Ashish Vaswani, Noa...

work page 2013
[33]

Graph Attention Networks

Graph attention networks.arXiv preprint arXiv:1710.10903. Devika Venugopalan, Ziwen Yan, Conrad Borchers, Jionghao Lin, and Vincent Aleven. 2025. Combining large language models with tutoring system intelli- gence: A case study in caregiver homework support. InProceedings of the 15th International Learning An- alytics and Knowledge Conference, pages 373–3...

work page internal anchor Pith review Pith/arXiv arXiv 2025
[34]

Charles J Wallace, Connie J Nelson, Robert Paul Liber- man, Robert A Aitchison, David Lukoff, John P El- der, and Chris Ferris

Psychadapter: adapting llms to reflect traits, personality, and mental health.npj Artificial Intelli- gence, 2(1):26. Charles J Wallace, Connie J Nelson, Robert Paul Liber- man, Robert A Aitchison, David Lukoff, John P El- der, and Chris Ferris. 1980. A review and critique of social skills training with schizophrenic patients. Schizophrenia Bulletin, 6(1)...

work page arXiv 1980
[35]

Mengxi Xiao, Qianqian Xie, Ziyan Kuang, Zhicheng Liu, Kailai Yang, Min Peng, Weiguang Han, and Jimin Huang

Teaching according to students’ aptitude: Personalized mathematics tutoring via persona-, memory-, and forgetting-aware llms.arXiv preprint arXiv:2511.15163. Mengxi Xiao, Qianqian Xie, Ziyan Kuang, Zhicheng Liu, Kailai Yang, Min Peng, Weiguang Han, and Jimin Huang. 2024. HealMe: Harnessing cognitive reframing in large language models for psychother- apy. ...

work page arXiv 2024
[36]

Hamad Zogan, Imran Razzak, Shoaib Jameel, and Guan- dong Xu

Investigating large language models in infer- ring personality traits from user conversations.arXiv preprint arXiv:2501.07532. Hamad Zogan, Imran Razzak, Shoaib Jameel, and Guan- dong Xu. 2021. Depressionnet: A novel summariza- tion boosted deep framework for depression detection on social media.arXiv preprint arXiv:2105.10878. 56

work page arXiv 2021