Elamur Research⌕
☰
home / papers / 2026 / MAQuA: Multi-outcome Adaptive Question-Asking for Mental Health using Item Response Theory

MAQuA: Multi-outcome Adaptive Question-Asking for Mental Health using Item Response Theory

Paper · 2026 · Vasudha Varadarajan, Hui Xu, Rebecca Astrid Böhme et al.

Summary

The paper tackles a practical problem in LLM-assisted mental health screening: asking too many questions tires users. Most existing language-based models also estimate only one condition at a time, even though comorbidity is common. The authors propose MAQuA, which pairs multi-task prediction of several mental health scores from free-text answers with multidimensional item response theory (MIRT) and exploratory factor analysis. At each turn, the system chooses the next question expected to give the most information across all dimensions together. This extends earlier single-dimension adaptive language-based assessment (ALBA).

To test it, the team built a new dataset that combines standard Likert-scale questionnaires with open-ended written responses from people reporting professional diagnoses. Multi-task models predicted ten outcomes more accurately than single-task baselines. Adaptive ordering reached stable scores with far fewer questions than random ordering, for both internalizing conditions (depression, anxiety) and externalizing ones (substance use, eating disorders). This points toward shorter, less burdensome screening that could work alongside LLM agents. The authors stress it is meant for screening, not diagnosis.

Key points

  • Dataset: 515 English-speaking participants (297 female, 182 male, 35 non-binary, 1 undisclosed), recruited mainly in the UK, Sweden and the US.
    • Participants had self-reported diagnoses across nine conditions, including AD, BD, depression, ADHD, PTSD, OCD, ED and addiction.
    • Comorbid cases were deliberately included; ADHD is underrepresented.
  • Compared with random question order, MAQuA cut the questions needed for score stabilization by 50–87%.
    • Depression needed 71% fewer questions.
    • Eating disorders needed 85% fewer questions.
  • The method combines multi-outcome language modeling, exploratory factor analysis and MIRT-based item selection, with early stopping rules.
    • Questions come from a fixed pool rather than being generated, which guards against LLM hallucinated questions.
  • Code is released at github.com/humanlab/MAQuA-IRT-framework; the dataset is available on request.
    • The study was approved by the Lund University IRB.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Lund UniversityLund · Sweden11 works · 1,263 citations🎓Vanderbilt UniversityNashville · United States14 works · 1,413 citations🎓Aarhus UniversityAarhus · Denmark6 works · 12 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksPsychology · 1,552 worksSocial Psychology · 306 worksApplied Psychology · 21 works

Cite

Cite as Vasudha Varadarajan, Hui Xu, Rebecca Astrid Böhme, Mariam Marlen Mirström, Sverker Sikström and H. Andrew Schwartz. 2026. MAQuA: Multi-outcome Adaptive Question-Asking for Mental Health using Item Response Theory. Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.eacl-long.313

@inproceedings{varadarajan2026maqua,
  title = {{MAQuA: Multi-outcome Adaptive Question-Asking for Mental Health using Item Response Theory}},
  author = {Vasudha Varadarajan, Hui Xu, Rebecca Astrid Böhme, Mariam Marlen Mirström, Sverker Sikström and H. Andrew Schwartz},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.eacl-long.313},
  url = {https://doi.org/10.18653/v1/2026.eacl-long.313}
}

References

53 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Lenard A Adler, Thomas Spencer, Stephen V Faraone, Ronald C Kessler, Mary J Howes, Joseph Biederman, and Kristina Secnik. 2006. Validity of pilot adult adhd self-report scale (asrs) to rate adult adhd symptoms. Annals of Clinical Psychiatry, 18(3):145–148. find
  2. John P Allen, Raye Z Litten, Joanne B Fertig, and Thomas Babor. 1997. A review of research on the alcohol use disorders identification test (audit). Alcoholism: clinical and experimental research, 21(4):613–619. find
  3. Emily Alsentzer, John R Murphy, Willie Boag, Wei-Hung Weng, Di Jin, Tristan Naumann, and Matthew McDermott. 2019. Publicly available clinical bert embeddings. arXiv preprint arXiv:1904.03323.
  4. Anne H Berman, Hans Bergman, Tom Palmstierna, and Frans Schlyter. 2007. Dudit. The Drug Use Disorders Identification Test–E. MANUAL. Karolinska institutet, Stockholm. find
  5. R Philip Chalmers. 2012. mirt: A multidimensional item response theory package for the r environment. Journal of statistical Software, 48:1–29. find
  6. R Philip Chalmers. 2016. Generating adaptive and non-adaptive test interfaces for multidimensional item response theory applications. Journal of Statistical Software, 71:1–38. find
  7. Glen Coppersmith, Mark Dredze, Craig Harman, Kristy Hollingshead, and Margaret Mitchell. 2015. find
  8. CLPsych 2015 Shared Task: Depression and PTSD on Twitter. In Proceedings of the 2nd Workshop on Computational Linguistics and Clinical Psychology: From Linguistic Signal to Clinical Reality, pages 31– 39. Association for Computational Linguistics.
  9. Robert Cudeck and Robert C MacCallum. 2000. Exploratory Factor Analysis. Lawrence Erlbaum Associates, Mahwah, NJ. find
  10. Johannes C Eichstaedt, Robert J Smith, Raina M Merchant, Lyle H Ungar, Patrick Crutchley, Daniel Preoţiuc-Pietro, David A Asch, and H Andrew Schwartz. 2018. Facebook language predicts depression in medical records. Proceedings of the National Academy of Sciences, 115(44):11203–11208. find
  11. Susan E Embretson and Steven P Reise. 2000. Item Response Theory for Psychologists. Lawrence Erlbaum Associates, Mahwah, NJ. find
  12. Jonna M Eriksson, Lisa MJ Andersen, and Susanne Be-jerot. 2013. Raads-14 screen: validity of a screening tool for autism spectrum disorder in an adult psychiatric population. Molecular Autism, 4:1–11. find
  13. Christopher G Fairburn and Sarah J Beglin. 2008. Eating disorder examination questionnaire. Cognitive behavior therapy and eating disorders, 309:313. find
  14. Adithya V Ganesan, Vasudha Varadarajan, Yash Kumar Lal, Veerle C Eijsbroek, Katarina Kjell, Oscar NE Kjell, Tanuja Dhanasekaran, Elizabeth C Stade, Johannes C Eichstaedt, Ryan L Boyd, and 1 others. 2024. Explaining gpt-4’s schema of depression using machine behavior analysis. arXiv preprint arXiv:24… — [arXiv:2411.13800](https://arxiv.org/abs/2411.13800) find
  15. Wayne K Goodman, Lawrence H Price, Steven A Rasmussen, Carolyn Mazure, Roberta L Fleischmann, Candy L Hill, George R Heninger, and Dennis S Charney. 1989. The yale-brown obsessive compulsive scale: I. development, use, and reliability. Archives of general psychiatry, 46(11):1006–1011. find
  16. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, and 1 others. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  17. Ronald K Hambleton, Hariharan Swaminathan, and H Jane Rogers. 1991. Fundamentals of Item Response Theory. Sage Publications, Newbury Park, CA. find
  18. Keith Harrigian, Carlos Aguirre, and Mark Dredze. 2020. Do models of mental health based on social media data generalize? In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 3774–3788. find
  19. Ian Andrew James, Rachel Morse, and Alan Howarth. 2010. The science and art of asking questions in cognitive therapy. Behavioural and Cognitive Psychotherapy, 38(1):83–93. find
  20. Shaoxiong Ji, Tianlin Zhang, Luna Ansari, Jie Fu, Prayag Tiwari, and Erik Cambria. 2022. Mental-BERT: Publicly Available Pretrained Language Models for Mental Healthcare. In Proceedings of the 13th Language Resources and Evaluation Conference (LREC), pages 7184–7190. find
  21. Seungwan Jin, Bogoan Kim, and Kyungsik Han. 2025. “I Don’t Know Why I Should Use This App”: Holistic Analysis on User Engagement Challenges in Mobile Mental Health. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, pages 1–23. find
  22. Swanie Juhng, Matthew Matero, Vasudha Varadarajan, Johannes Eichstaedt, Adithya V Ganesan, and H Andrew Schwartz. 2023. Discourse-level representations can improve prediction of degree of anxiety. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: S… find
  23. Oscar NE Kjell, Salvatore Giorgi, H Andrew Schwartz, Lyle H Ungar, David B Yaden, Margaret L Kern, and Johannes C Eichstaedt. 2022. Natural language analyzed with AI-based transformers predict traditional subjective well-being measures approaching the theoretical upper limits in accuracy. Scientific… find
  24. Kurt Kroenke, Robert L Spitzer, and Janet BW Williams. 2001. The PHQ-9: validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9):606–613. find
  25. Lorenz Kuhn, Yarin Gal, and Sebastian Farquhar. 2023. Clam: Selective clarification for ambiguous questions with generative language models. In Workshop on Challenges in Deployable Generative AI at International Conference on Machine Learning (ICML). ArXiv preprint arXiv:2212.07769. find
  26. Aditya Kusupati, Gantavya Bhatt, Aniket Rege, Matthew Wallingford, Aditya Sinha, Vivek Ramanujan, William Howard-Snyder, Kaifeng Chen, Sham Kakade, Prateek Jain, and Ali Farhadi. 2022. Matryoshka representation learning. In Advances in Neural Information Processing Systems, volume 35, pages 30233–30… find
  27. John L Horn. 1965. A rationale and test for the number of factors in factor analysis. Psychometrika, 30(2):179–185. find
  28. John P. Lalor, Hao Wu, and Hong Yu. 2016. Building an evaluation scale using item response theory. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 648– 657, Austin, Texas. Association for Computational Linguistics. find
  29. Jihyun K Hur, Joseph Heffner, Gloria W Feng, Jutta Joormann, and Robb B Rutledge. 2024. Language sentiment predicts changes in depressive symptoms. Proceedings of the National Academy of Sciences, 121(39):e2321321121. find
  30. Richard LeBeau, Emily Mischel, Heidi Resnick, Dean Kilpatrick, Matthew Friedman, and Michelle Craske. 2014. Dimensional assessment of posttraumatic stress disorder in dsm-5. Psychiatry research, 218(1-2):143–147. find
  31. Sean Lee, Aamir Shakir, Darius Koenig, and Julius Lipp. 2024. Open source strikes bread - new fluffy embeddings model. find
  32. Shuyue Stella Li, Jimin Mun, Faeze Brahman, Jonathan S Ilgen, Yulia Tsvetkov, and Maarten Sap. 2025. Aligning llms to ask good questions a case study in clinical reasoning. arXiv preprint arXiv:2502.14860. — [arXiv:2502.14860](https://arxiv.org/abs/2502.14860) find
  33. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining approach. Preprint, arXiv:1907.11692.
  34. Syeda Mahwish, Ryan L. Boyd, Vasudha Varadarajan, Roman Kotov, Benjamin J. Luft, H. Andrew Schwartz, and Sean A. P. Clouston. 2026. Measuring resilience using language modeling: A computational approach to observing resilience. Journal of Traumatic Stress, pages 1–14. find
  35. Mark D Reckase. 2006. 18 multidimensional item response theory. Handbook of statistics, 26:607–642. find
  36. Gony Rosenman, Talma Hendler, and Lior Wolf. 2024. LLM questionnaire completion for automatic psychiatric assessment. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 403–415, Miami, Florida, USA. Association for Computational Linguistics. find
  37. Ozan Sener and Vladlen Koltun. 2018. Multi-task learning as multi-objective optimization. In Advances in Neural Information Processing Systems, volume 31. find
  38. Chen Shani and Elizabeth Stade. 2025. Measuring mental health variables in computational research: Toward validated, dimensional, and transdiagnostic approaches. In Proceedings of the 10th Workshop on Computational Linguistics and Clinical Psychology (CLPsych 2025), pages 69–78, Albuquerque, New Mex… find
  39. Rosmarie Mendel, Eva Traut-Mattausch, Eva Jonas, Stefan Leucht, John M Kane, Katja Maino, Werner Kissling, and Johannes Hamann. 2011. Confirmation bias: why psychiatrists stick to wrong preliminary diagnoses. Psychological medicine, 41(12):2651– 2659. find
  40. Guangyao Shen, Jia Jia, Liqiang Nie, Fuli Feng, Cunjun Zhang, Tianrui Hu, Tat-Seng Chua, and Wenwu Zhu. 2017. Depression detection via harvesting social media: A multimodal dictionary learning solution. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJC… find
  41. Christopher J Miller, Joshua Klugman, Debra A Berv, Kristen J Rosenquist, and S Nassir Ghaemi. 2004. The sensitivity and specificity of the Mood Disorder Questionnaire for detecting bipolar disorder. Journal of Affective Disorders, 81(2):167–171. find
  42. Sverker Sikström, Oscar NE Kjell, Katarina Kjell, and Johannes Lundberg. 2023. Precise language models can estimate psychological states and abilities from text. Scientific Reports, 13(1):11679. find
  43. Birger Moëll. 2024. Comparing the efficacy of gpt-4 and chat-gpt in mental health care: A blind assessment of large language models for psychological support. arXiv preprint arXiv:2405.09300. — [arXiv:2405.09300](https://arxiv.org/abs/2405.09300) find
  44. Viet Cuong Nguyen, Mohammad Taher, Dongwan Hong, Vinicius Konkolics Possobom, Vibha Thirunel-layi Gopalakrishnan, Ekta Raj, Zihang Li, Heather J. Soled, Michael L. Birnbaum, Srijan Kumar, and Munmun De Choudhury. 2025. Do large language models align with core mental health counseling competencies? I… find
  45. Zach Nussbaum, John X Morris, Brandon Duderstadt, and Andriy Mulyar. 2024. Nomic embed: Training a reproducible long context text embedder. arXiv preprint arXiv:2402.01613.
  46. OpenAI. 2023. GPT-4 Technical Report. https://openai.com/research/gpt-4. ↗
  47. David Owen, Jose Camacho-Collados, and Luis Espinosa Anke. 2020. Towards preemptive detection of depression and anxiety in Twitter. In Proceedings of the Fifth Social Media Mining for Health Applications Workshop & Shared Task, pages 82–89. Association for Computational Linguistics. find
  48. Khushboo Singh, Vasudha Varadarajan, Adithya V Ganesan, August Håkan Nilsson, Nikita Soni, Syeda Mahwish, Pranav Chitale, Ryan L. Boyd, Lyle Ungar, Richard N Rosenthal, and H. Schwartz. 2025. Systematic evaluation of auto-encoding and large language model representations for capturing author states… find
  49. Nikita Soni, August Håkan Nilsson, Syeda Mahwish, Vasudha Varadarajan, H. Andrew Schwartz, and Ryan L. Boyd. 2025. Who we are, where we are: Mental health at the intersection of person, situation, and large language models. In Proceedings of the 10th Workshop on Computational Linguistics and Clinic… find
  50. Robert L Spitzer, Kurt Kroenke, Janet BW Williams, and Bernd Löwe. 2006. A brief measure for assessing generalized anxiety disorder: the gad-7. Archives of internal medicine, 166(10):1092–1097. find
  51. Vasudha Varadarajan, Allison Lahnala, Adithya V Ganesan, Gourab Dey, Siddharth Mangalik, Ana-Maria Bucur, Nikita Soni, Rajath Rao, Kevin Lanning, Isabella Vallejo, Lucie Flek, H. Andrew Schwartz, Charles Welch, and Ryan Boyd. 2024a. Archetypes and entropy: Theory-driven extraction of evidence for su… find
  52. Vasudha Varadarajan, Allison Lahnala, Sujeeth Vanku-dari, Akshay Raghavan, Scott Feltman, Syeda Mahwish, Camilo Ruggero, Roman Kotov, and H. Andrew Schwartz. 2025. Linking language-based distortion detection to mental health outcomes. In Proceedings of the 10th Workshop on Computational Linguistics… find
  53. Vasudha Varadarajan, Sverker Sikström, Oscar Kjell, and H. Andrew Schwartz. 2024b. ALBA: Adaptive Language-Based Assessments for Mental Health. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/varadarajan-2026-maqua-multi-outcome-adaptive-question/