Elamur Research⌕
☰
home / papers / 2026 / MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric Diagnosis

MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric Diagnosis

Paper · 2026 · Xiao Sun, Ymyang, Xinyi Jiang et al.

Summary

The paper tackles automated psychiatric diagnosis with LLMs. Existing evaluation data are mostly social-media posts or synthetic stories with coarse labels, so they say little about real clinical use. The authors build MentalDx Bench from 712 de-identified electronic health records collected at partner hospitals. Psychiatrists labelled every record under ICD-11 at two levels: 16 broad categories and 76 specific disorders.

They tested 18 LLMs on it. The models usually identified the broad category but often failed to name the exact disorder. The authors trace this to keyword-to-label shortcuts, whereas clinicians reason by forming and ruling out hypotheses.

To close this gap they propose MentalSeek-Dx, built on Qwen2.5. Training has two stages. First, supervised fine-tuning on reasoning traces that sort symptoms, propose candidate diagnoses and run differential exclusion. Second, curriculum reinforcement learning that moves from category-level to disorder-level rewards.

The 14B model achieves state-of-the-art accuracy, beating much larger general-purpose systems. This suggests that aligning a model's reasoning with clinical practice matters more than model scale.

Key points

  • MentalDx Bench data: 712 EHRs labelled with ICD-11 Chapter 6A names and codes, covering 76 disorders in 16 categories. Comorbid and incomplete cases were excluded, and sampling was balanced across labels.
  • Annotation: 12 board-certified psychiatrists took part. Each case received two independent senior labels with Cohen's κ > 0.8, and disagreements were settled by consensus.
  • Evaluation finding: across 18 LLMs, category-level accuracy was respectable while disorder-level accuracy collapsed, which the authors call a "paradigm misalignment".
  • Ablation on Qwen2.5-14B: adding hypothetico-deductive trajectories raised category accuracy (CA) by 13.88%, disorder accuracy (DA) by 20.22% and joint accuracy (JA) by 22.36% compared with direct training.
  • Effect of RL on Qwen2.5-7B: RL added 3.23% DA and 3.37% JA. The curriculum gave modest accuracy gains but mainly stabilised training.

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

🎓Tsinghua UniversityBeijing · China333 works · 87,019 citations🏛️Chinese Academy of SciencesBeijing · China194 works · 23,906 citations🔬Institute of AutomationBeijing · China70 works · 14,251 citations🎓UNSW SydneySydney · Australia34 works · 1,509 citations🎓Chongqing UniversityChongqing · China7 works · 318 citations

Also on the byline: The Affiliated Yongchuan Hospital of Chongqing Medical University.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksPsychology · 1,552 worksSocial Psychology · 306 worksHealth Professions · 69 worksHealth Information Management · 50 works

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 1,426 works

Cite

Cite as Xiao Sun, Ymyang, Xinyi Jiang, Yu Tian, Junnan Zhu, Jiang Zhong, Qin Lei, Jingwang Huang, Haoyang Zeng, Xinyu Zhou, Xin Xiao and Kaiwen Wei. 2026. MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric Diagnosis. Proceedings of the 64th Annual Meeting 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.acl-long.1225

@inproceedings{sun2026mentalseek,
  title = {{MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric Diagnosis}},
  author = {Xiao Sun, Ymyang, Xinyi Jiang, Yu Tian, Junnan Zhu, Jiang Zhong, Qin Lei, Jingwang Huang, Haoyang Zeng, Xinyu Zhou, Xin Xiao and Kaiwen Wei},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1225},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1225}
}

References

52 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sandhini Agarwal, Lama Ahmad, Jason Ai, Sam Altman, Andy Applebaum, Edwin Arbus, Rahul K Arora, Yu Bai, Bowen Baker, Haiming Bao, and 1 others. 2025. gpt-oss-120b & gpt-oss-20b model card. arXiv preprint arXiv:2508.10925.
  2. Ruth Appel, Peter McCrory, Alex Tamkin, Miles McCain, Tyler Neylon, and Michael Stern. 2025. Anthropic economic index report: Uneven geographic and enterprise ai adoption. arXiv preprint arXiv:2511.15080. — [arXiv:2511.15080](https://arxiv.org/abs/2511.15080) find
  3. Eliseo Bao, Anxo Pérez, and Javier Parapar. 2025. Redsm5: A reddit dataset for dsm-5 depression detection. In Proceedings of the 34th ACM International Conference on Information and Knowledge Management, pages 6323–6327. find
  4. Sébastien Bubeck, Christian Coester, Ronen Eldan, Timothy Gowers, Yin Tat Lee, Alexandru Lupsasca, Mehtaab Sawhney, Robert Scherrer, Mark Sellke, Brian K Spears, and 1 others. 2025. Early science acceleration experiments with gpt-5. arXiv preprint arXiv:2511.16072. — [arXiv:2511.16072](https://arxiv.org/abs/2511.16072) find
  5. Grace Byun, Rebecca Lipschutz, Sean T Minton, Abigail Lott, and Jinho D Choi. 2025. Cradle bench: A clinician-annotated benchmark for multi-faceted mental health crisis and safety risk detection. arXiv preprint arXiv:2510.23845. find
  6. Junying Chen, Zhenyang Cai, Ke Ji, Xidong Wang, Wanlong Liu, Rongsheng Wang, Jianye Hou, and Benyou Wang. 2024. Huatuogpt-o1, towards medical complex reasoning with llms. Preprint, arXiv:2412.18925. find
  7. Arman Cohan, Bart Desmet, Andrew Yates, Luca Soldaini, Sean MacAvaney, and Nazli Goharian. 2018. Smhd: a large-scale resource for exploring online language usage for multiple mental health conditions. arXiv preprint arXiv:1806.05258.
  8. Mark Collins. 2017. Clinical reasoning in image guided radiotherapy: a multimethod study. Sheffield Hallam University (United Kingdom). find
  9. Chongyuan Dai, Jinpeng Hu, Hongchang Shi, Zhuo Li, Xun Yang, and Meng Wang. 2025. Psyche-r1: Towards reliable psychological llms through unified empathy, expertise, and reasoning. arXiv preprint arXiv:2508.10848.
  10. Chengfeng Dou, Chong Liu, Fan Yang, Fei Li, Jiyuan Jia, Mingyang Chen, Qiang Ju, Shuai Wang, Shunya Dang, Tianpeng Li, and 1 others. 2025. Baichuanm2: Scaling medical capability with large verifier system. arXiv preprint arXiv:2509.02208.
  11. Aya E Fouda, Abdelrahamn A Hassan, Radwa J Hanafy, and Mohammed E Fouda. 2025. Psychiatrybench: A multi-task benchmark for llms in psychiatry. arXiv preprint arXiv:2509.09711. — [arXiv:2509.09711](https://arxiv.org/abs/2509.09711) find
  12. Clara Frydman-Gani, Alejandro Arias, Maria Perez Vallejo, John Daniel Londoño Martínez, Johanna Valencia-Echeverry, Mauricio Castaño, Alex AT Bui, Nelson B Freimer, Carlos Lopez-Jaramillo, and Loes M Olde Loohuis. 2025. Large language models for psychiatric phenotype extraction from electronic healt… find
  13. Changzeng Fu, Zelin Fu, Qi Zhang, Xinhe Kuang, Jiacheng Dong, Kaifeng Su, Yikai Su, Wenbo Shi, Junfeng Yao, Yuliang Zhao, and 1 others. 2025. The first mpdd challenge: Multimodal personality-aware depression detection. arXiv preprint arXiv:2505.10034. — [arXiv:2505.10034](https://arxiv.org/abs/2505.10034) find
  14. Zhuohan Ge, Nicole Hu, Darian Li, Yubo Wang, Shihao Qi, Yuming Xu, Han Shi, and Jason Zhang. 2025. A survey of large language models in mental health disorder detection on social media. In 2025 IEEE 41st International Conference on Data Engineering Workshops (ICDEW), pages 164–176. IEEE. find
  15. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948.
  16. Amey Hengle, Atharva Kulkarni, Shantanu Deepak Patankar, Madhumitha Chandrasekaran, Sneha D’silva, Jemima S Jacob, and Rashmi Gupta. 2024. Still not quite there! evaluating large language models for comorbid mental health diagnosis. In Proceedings of the 2024 Conference on Empirical Methods in Natur… find
  17. Jinpeng Hu, Tengteng Dong, Gang Luo, Hui Ma, Peng Zou, Xiao Sun, Dan Guo, Xun Yang, and Meng Wang. 2024. Psycollm: Enhancing llm for psychological understanding and evaluation. IEEE Transactions on Computational Social Systems, 12(2):539–551. find
  18. Mariia Ignashina, Paulina Bondaronek, Dan Santel, John Pestian, and Julia Ive. 2025. Llm assistance for pediatric depression. arXiv preprint arXiv:2501.17510. — [arXiv:2501.17510](https://arxiv.org/abs/2501.17510) find
  19. Elaine Jefford, Kathleen Fahy, and Deborah Sundin. 2011. Decision-making theories and their usefulness to the midwifery profession both in terms of midwifery practice and the education of midwives. International Journal of Nursing Practice, 17(3):246– 253. find
  20. Shreya Johri, Jaehwan Jeong, Benjamin A Tran, Daniel I Schlessinger, Shannon Wongvibulsin, Leandra A Barnes, Hong-Yu Zhou, Zhuo Ran Cai, Eliezer M Van Allen, David Kim, and 1 others. 2025. An evaluation framework for clinical use of large language models in patient interaction tasks. Nature medicine… find
  21. Laerdon Kim. 2025. A baseline for self-state identification and classification in mental health data: Clpsych 2025 task. arXiv preprint arXiv:2504.14066. find
  22. Wuyang Lan, Wenzheng Wang, Changwei Ji, Guoxing Yang, Yongbo Zhang, Xiaohong Liu, Song Wu, and Guangyu Wang. 2025a. Clinicalgpt-r1: Pushing reasoning capability of generalist disease diagnosis with large language model. arXiv preprint arXiv:2504.09421. — [arXiv:2504.09421](https://arxiv.org/abs/2504.09421) find
  23. Xiaochong Lan, Zhiguang Han, Yiming Cheng, Li Sheng, Jie Feng, Chen Gao, and Yong Li. 2025b. Depression detection on social media with large language models. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 2155–2171. find
  24. Neil Langridge, Lisa Roberts, and Catherine Pope. 2016. The role of clinician emotion in clinical reasoning: Balancing the analytical process. Manual therapy, 21:277–281. find
  25. Inbar Levkovich, Eyal Rabin, Michal Brann, and Zohar Elyoseph. 2024. Large language models outperform general practitioners in identifying complex cases of childhood anxiety. Digital Health, 10:20552076241294182. find
  26. Silvia Kyungjin Lho, Sang-Cheol Park, Hahyun Lee, Da Young Oh, Hyeonjin Kim, Soomin Jang, Hee Yeon Jung, So Young Yoo, Su Mi Park, and Jun-Young Lee. 2025. Large language models and text embeddings for detecting depression and suicide in patient narratives. JAMA Network Open, 8(5):e2511922–e2511922. find
  27. Fenglin Liu, Shen Ge, and Xian Wu. 2021. Competence-based multimodal curriculum learning for medical report generation. In Proceedings of the 59th annual meeting of the association for computational linguistics and the 11th international joint conference on natural language processing (volume 1: lon… find
  28. Shuyu Liu, Ruoxi Wang, Ling Zhang, Xuequan Zhu, Rui Yang, Xinzhu Zhou, Fei Wu, Zhi Yang, Cheng Jin, and Gang Wang. 2025. Psychbench: A comprehensive and professional benchmark for evaluating the performance of llm-assisted psychiatric clinical practice. arXiv preprint arXiv:2503.01903. — [arXiv:2503.01903](https://arxiv.org/abs/2503.01903) find
  29. Siddharth Mangalik, Johannes C Eichstaedt, Salvatore Giorgi, Jihu Mun, Farhan Ahmed, Gilvir Gill, Adithya V. Ganesan, Shashanka Subrahmanya, Nikita Soni, Sean AP Clouston, and 1 others. 2024. Robust language-based mental health assessments in time and space through social media. NPJ Digital Medicine… find
  30. Saad Mankarious, Ayah Zirikly, Daniel Wiechmann, Elma Kerz, Edward Kempa, and Yu Qiao. 2025. Mindset: Advancing mental health benchmarking through large-scale social media data. arXiv preprint arXiv:2511.20672.
  31. Nick Obradovich, Sahib S Khalsa, Waqas U Khan, Jina Suh, Roy H Perlis, Olusola Ajilore, and Martin P Paulus. 2024. Opportunities and risks of large language models in psychiatry. NPP—Digital Psychiatry and Neuroscience, 2(1):8. find
  32. Jaedong Oh, Jooyoung Lim, and Hayoung Oh. 2026. Clinically validated depression dataset aligned with dsm-5 criteria for major depressive disorder (mdd). Expert Systems with Applications, 296. find
  33. Graziella Orrù, Giulia Melis, and Giuseppe Sartori. 2025. Large language models and psychiatry. International Journal of Law and Psychiatry, 101:102086. find
  34. David Patterson, Joseph Gonzalez, Urs Hölzle, Quoc Le, Chen Liang, Lluis-Miquel Munguia, Daniel Rothchild, David R So, Maud Texier, and Jeff Dean. 2022. The carbon footprint of machine learning training will plateau, then shrink. Computer, 55(7):18– 28. find
  35. Ahmad Radwan, Mohannad Amarneh, Hussam Alawneh, Huthaifa I Ashqar, Anas AlSobeh, and Aws Abed Al Raheem Magableh. 2024. Predictive analytics in mental health leveraging llm embeddings and machine learning models for social media analysis. International Journal of Web Services Research (IJWSR), 21(1)… find
  36. Mavin Sao and Hoi-Jeong Lim. 2024. Miroberta: Mental illness text classification with transfer learning on subreddits. IEEE Access. find
  37. Andrew Sellergren, Sahar Kazemzadeh, Tiam Jaroensri, Atilla Kiraly, Madeleine Traverse, Timo Kohlberger, Shawn Xu, Fayaz Jamil, Cían Hughes, Charles Lau, and 1 others. 2025. Medgemma technical report. arXiv preprint arXiv:2507.05201. find
  38. Ryan Allen Shewcraft, John Schwarz, and Mariann Micsinai Balan. 2025. Algorithmic classification of psychiatric disorder–related spontaneous communication using large language model embeddings: Algorithm development and validation. JMIR AI, 4:e67369. find
  39. Daun Shin, Hyoseung Kim, Seunghwan Lee, Younhee Cho, and Whanbo Jung. 2024. Using large language models to detect depression from user-generated diary text data as a novel approach in digital mental health screening: instrument validation study. Journal of Medical Internet Research, 26:e54617. find
  40. Kimi Team, Yifan Bai, Yiping Bao, Guanduo Chen, Jiahao Chen, Ningxin Chen, Ruijue Chen, Yanru Chen, Yuankun Chen, Yutian Chen, and 1 others. 2025. Kimi k2: Open agentic intelligence. arXiv preprint arXiv:2507.20534.
  41. Qwen Team and 1 others. 2024. Qwen2 technical report. arXiv preprint arXiv:2407.10671, 2(3). find
  42. Xin Wang, Yudong Chen, and Wenwu Zhu. 2021. A survey on curriculum learning. IEEE transactions on pattern analysis and machine intelligence, 44(9):4555–4576. find
  43. Yi-Fan Wang, Ming-Da Li, Su-Hong Wang, Yin Fang, Jie Sun, Lin Lu, and Wei Yan. 2025. Large language models in clinical psychiatry: Applications and optimization strategies. World Journal of Psychiatry, 15(11):108199. find
  44. Samantha Weber, Nicolas Deperrois, Robert Heun, Laura Frühschütz, Anna Monn, Stephanie Homan, Andrea Häfliger, Erich Seifritz, Tobias Kowatsch, MULTICAST consortium Jäger Lena 8 Schulte-braucks Katharina 9 Gershov Sapir 9 Mocellin Jacopo 1 4, and 1 others. 2025. Using a fine-tuned large language mod… find
  45. Shurui Wu, Xinyi Huang, and Dingxin Lu. 2025. Psychological health knowledge-enhanced llm-based social network crisis intervention text transfer recognition method. In Proceedings of the 2025 International Conference on Health Big Data, pages 156– 161. find
  46. Xuhai Xu, Bingsheng Yao, Yuanzhe Dong, Saadia Gabriel, Hong Yu, James Hendler, Marzyeh Ghassemi, Anind K Dey, and Dakuo Wang. 2024. Mental-llm: Leveraging large language models for mental health prediction via online text data. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous T… find
  47. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, and 1 others. 2025. Qwen3 technical report. arXiv preprint arXiv:2505.09388.
  48. Kailai Yang, Tianlin Zhang, Ziyan Kuang, Qianqian Xie, Jimin Huang, and Sophia Ananiadou. 2024. Mental-lama: interpretable mental health analysis on social media with large language models. In Proceedings of the ACM Web Conference 2024, pages 4489–4500. find
  49. Binwei Yao, Chao Shi, Likai Zou, Lingfeng Dai, Mengyue Wu, Lu Chen, Zhen Wang, and Kai Yu. 2022. D4: a chinese dialogue dataset for depression-diagnosis-oriented chat. arXiv preprint arXiv:2205.11764. find
  50. Shouwen Zheng, Yingzhi Tao, and Taiqi Zhou. 2025. Rsd-15k: A large-scale user-level annotated dataset for suicide risk detection on social media. In 2025 IEEE 41st International Conference on Data Engineering Workshops (ICDEW), pages 190–196. IEEE. find
  51. Mingcheng Zhu, Zhiyao Luo, Yu Liu, and Tingting Zhu. 2025. Medtpe: Compressing long ehr sequence for llm-based clinical prediction with token-pair encoding. In Proceedings of the 11th Mining and Learning from Time Series Workshop@ KDD, volume 2025. find
  52. Muni Zhuang, Dongsheng Cheng, Xin Lu, and Xu Tan. 2024. Postgraduate psychological stress detection from social media using bert-fused model. PloS one, 19(10):e0312264. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/sun-2026-mentalseek-dx-progressive-hypothetico-deductive/