Elamur Research⌕
☰
home / papers / 2026 / McMasters of Change: Predicting Well-Being States and Transitions from Longitudinal Language

McMasters of Change: Predicting Well-Being States and Transitions from Longitudinal Language

Paper · 2026 · Hongyi Zhang, Derron Li, Scarlett Cleary et al.

Summary

Most mental-health NLP treats social media posts one at a time and ignores how a person's state unfolds over a timeline. This paper describes McMaster NLP's entry to the CLPsych 2026 Shared Task, which is built on the MIND framework and has three tasks. The first is detecting adaptive and maladaptive ABCD self-state subelements and their presence. The second is spotting moments of change, both switches and escalations. The third is producing structured summaries.

For Task 1, each post combines mental-roberta-base embeddings with cosine similarities to 32 ABCD archetype sentences. ChatGPT 5.1 wrote the archetypes, all-MiniLM-L6-v2 encodes them, and per-target linear models make the predictions. Task 2 uses a Ridge wellbeing regressor with a difference threshold for switches and a BiLSTM for escalations. Task 3 prompts Qwen3.5-9B without fine-tuning.

The system beat the baselines on Task 1, was mixed on Task 2, and produced consistent summaries with low content coverage. Error analysis shows that the models overpredict adaptive states and treat style, such as verbosity, as a sign of wellbeing. The authors also question the GAF-based wellbeing labels. Overall, the work shows that cheap LLM-written archetypes can bring interpretable gains without clinical expertise.

Key points

  • Task 1 features and data: Each post gets an 800-dimensional vector: a 768-d MentalRoBERTa embedding plus 32 archetype-similarity scores. Training data came from 30 labeled timelines. ModernBERT-based masked-token augmentation grew it from 312 to 495 posts (+58.7%).
  • Task 1 results: Both subtasks ranked seventh.
    • Subtask 1.1: Avg Sub Macro F1 rose from the baseline's 0.247 to 0.351, and adaptive subelement F1 rose from 0.156 to 0.317.
    • Subtask 1.2: RMSE dropped to 0.957 for maladaptive and 1.112 for adaptive presence. The baseline still had slightly better QWK and Spearman scores.
  • Task 2 method and results: A post counts as a switch when the predicted wellbeing differs by at least 2 from the previous post. The system ranked eleventh with a combined Macro F1 of 0.412. That is above both Llama zero-shot baselines but below TempoFormer. At the timeline level, escalation F1 was 0.660 and switch F1 was 0.274.
  • Task 3 results: Task 3.1 ranked ninth.
    • Consistency and contradiction scores were strong (CS 0.770, CT 0.761).
    • Content overlap was weak (ROUGE-L Recall 0.208, BERTScore Recall 0.255).
  • Proposed future work: The authors suggest moving beyond the GAF scale, for example to WHODAS-2 or the WHO-5 index. Any such instrument would need adapting to social media, and clinicians should help validate the archetypes.

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

🎓Carnegie Mellon UniversityPittsburgh · United States539 works · 179,533 citations🎓Indiana University BloomingtonBloomington · United States37 works · 3,685 citations🎓Indiana UniversityBloomington · United States41 works · 3,663 citations🎓McMaster UniversityHamilton · Canada4 works · 656 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksPsychology · 1,552 worksSocial Psychology · 306 works

Cite

Cite as Hongyi Zhang, Derron Li, Scarlett Cleary, Aadi Sanghani, Akshay Krishna Sirigana, Brian Miguel Pimentel, Kelsey Isman, Kian Omoomi, Vasudha Varadarajan, Charles Welch and Allison Lahnala. 2026. McMasters of Change: Predicting Well-Being States and Transitions from Longitudinal Language. Proceedings of the 11th Workshop on Computational Linguistics and Clinical Psychology (CLPsych 2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.clpsych-1.38

@inproceedings{zhang2026mcmasters,
  title = {{McMasters of Change: Predicting Well-Being States and Transitions from Longitudinal Language}},
  author = {Hongyi Zhang, Derron Li, Scarlett Cleary, Aadi Sanghani, Akshay Krishna Sirigana, Brian Miguel Pimentel, Kelsey Isman, Kian Omoomi, Vasudha Varadarajan, Charles Welch and Allison Lahnala},
  booktitle = {Proceedings of the 11th Workshop on Computational Linguistics and Clinical Psychology (CLPsych 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.clpsych-1.38},
  url = {https://doi.org/10.18653/v1/2026.clpsych-1.38}
}

References

26 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Substance Abuse and Mental Health Services Administration. 2016. Dsm-5 changes: Implications for child serious emotional disturbance [internet]. find
  2. Iqra Ali, Talia Tseriotou, Guy Dvir, Callum Chan, Yuxiang Zhou, Juan Antonio Lossio-Ventura, Ayal Klein, Aya Shamir, Dan Sayda, Anthony Hills, Aya Zirikly, Diana Inkpen, Dana Atzil-Slonim, and Maria Liakata. 2026. Overview of the clpsych 2026 shared task: Capturing and characterizing mental health c… find
  3. Dana Atzil-Slonim. 2024. Self-Other Dynamics (SOD): A Transtheoretical Coding Manual. Bar-Ilan University. Transtheoretical coding manual for psychotherapy research. find
  4. Dana Atzil-Slonim. 2025. Multimodal intrapersonal and interpersonal dynamics (mind): A transtheoretical coding manual. Open Science Framework. find
  5. Dana Atzil-Slonim. 2026. Leveraging theoretical and technological innovations to study the mechanisms that underlie therapeutic change in psychotherapy. In Louis G. Castonguay, Dana Atzil-Slonim, Michael Barkham, and Wolfgang Lutz, editors, Practice-Based Evidence in the Psychological Therapies: Tow… find
  6. Stevie Chancellor and Munmun De Choudhury. 2020. Methods in predictive techniques for mental health status on social media: a critical review. npj Digital Medicine, 3(1):43. find
  7. Sarah Clement, Oliver Schauman, Tanya Graham, Francesca Maggioni, Sara Evans-Lacko, Nikita Bezborodovs, Craig Morgan, Nicolas Rüsch, June SL Brown, and Graham Thornicroft. 2015. What is the impact of mental health-related stigma on help-seeking? a systematic review of quantitative and qualitative st… find
  8. Benjamin Lê Cook, Nhi-Ha Trinh, Zhihui Li, Sherry Shu-Yeu Hou, and Ana M Progovac. 2017. Trends in racial-ethnic disparities in access to mental health care, 2004–2012. Psychiatric services, 68(1):9–16. find
  9. Nils Reimers and Iryna Gurevych. 2019. Sentence-BERT: Sentence embeddings using Siamese BERT-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3982–3992,…
  10. Mike Schuster and Kuldip K. Paliwal. 1997. Bidirectional Recurrent Neural Networks. IEEE Transactions on Signal Processing, 45(11):2673–2681. find
  11. Dan J. Stein, Peter Szatmari, Wolfgang Gaebel, Michael Berk, Eduard Vieta, Mario Maj, Ymkje Anna De Vries, Annelieke M. Roest, Peter De Jonge, Andreas Maercker, Chris R. Brewin, Kathleen M. Pike, Carlos M. Grilo, Naomi A. Fineberg, Peer Briken, Peggy T. Cohen-Kettenis, and Geoffrey M. Reed. 2020. Me… find
  12. Esther MV Grootenboer, Erik J Giltay, Rosalind van der Lem, Tineke van Veen, Nic JA van der Wee, and Frans G Zitman. 2012. Reliability and validity of the global assessment of functioning scale in clinical outpatients with depressive disorders. Journal of evaluation in clinical practice, 18(2):502–5… find
  13. Scott Gspandl, Ryan P Peirson, Ramzi W Nahhas, Tracey Goodman Skale, and Douglas S Lehrer. 2018. Comparing global assessment of functioning (gaf) and world health organization disability assessment schedule (whodas) 2.0 in schizophrenia. Psychiatry Research, 259:251–253. find
  14. Molly E Ireland and James W Pennebaker. 2010. Language style matching in writing: Synchrony in essays, correspondence, and poetry. Journal of personality and social psychology, 99(3):549. find
  15. 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 Thirteenth Language Resources and Evaluation Conference, pages 7184–7190, Marseille, France. European Langu… find
  16. Huijie Lin, Jia Jia, Jiezhong Qiu, Yongfeng Zhang, Guangyao Shen, Lexing Xie, Jie Tang, Ling Feng, and Tat-Seng Chua. 2017. Detecting stress based on social interactions in social networks. IEEE Transactions on Knowledge and Data Engineering, 29(9):1820–1833. find
  17. Syeda Mahwish, Ryan L Boyd, Vasudha Varadarajan, Roman Kotov, Benjamin J Luft, H Andrew Schwartz, and Sean AP Clouston. 2026. Measuring resilience using language modeling: A computational approach to observing resilience. Journal of Traumatic Stress. find
  18. Harry L Piersma and Janna L Boes. 1997. The gaf and psychiatric outcome: a descriptive report. Community Mental Health Journal, 33(1):35–41. find
  19. Nils Reimers. 2021. sentence-transformers/all-MiniLM-L6-v2. https://huggingface.co/ sentence-transformers/all-MiniLM-L6-v2. Accessed: March 2026. ↗
  20. Qwen Team. 2026. Qwen3.5-omni technical report.
  21. Christian Winther Topp, Søren Dinesen Østergaard, Susan Søndergaard, and Per Bech. 2015. The who-5 well-being index: a systematic review of the literature. Psychotherapy and psychosomatics, 84(3):167– 176. find
  22. Adam Tsakalidis, Jenny Chim, Iman Munire Bilal, Ayah Zirikly, Dana Atzil-Slonim, Federico Nanni, Philip Resnik, Manas Gaur, Kaushik Roy, Becky Inkster, Jeff Leintz, and Maria Liakata. 2022. Overview of the CLPsych 2022 shared task: Capturing moments of change in longitudinal user posts. In Proceedin… find
  23. Talia Tseriotou, Jenny Chim, Ayal Klein, Aya Shamir, Guy Dvir, Iqra Ali, Cian Kennedy, Guneet Singh Kohli, Anthony Hills, Ayah Zirikly, Dana Atzil-Slonim, and Maria Liakata. 2025. Overview of the CLPsych 2025 shared task: Capturing mental health dynamics from social media timelines. In Proceedings o… find
  24. 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. 2024. Archetypes and entropy: Theory-driven extraction of evidence for sui… find
  25. Torbjørn Vatnaland, J Vatnaland, Svein Friis, and Stein Opjordsmoen. 2007. Are gaf scores reliable in routine clinical use? Acta Psychiatrica Scandinavica, 115(4):326–330. find
  26. World Health Organization. 2025. World mental health today: latest data. Technical report, World Health Organization, Geneva. License: CC BY-NC-SA 3.0 IGO. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhang-2026-mcmasters-change-predicting-well-being/