Validation And Ethics of the use of Large Language Models (Llms) as Affective Simulation Models in Psychopathology Research
Keywords:
large language model, agreement validity, affective simulation, artificial intelligence ethics, psychopathology researchAbstract
Large Language Models (LLMs) are increasingly used as affective simulation models in psychopathology research, yet their psychometric validity and safety-related ethical implications remain underexamined. This study evaluated agreement between affective severity scores generated by three LLMs and human reference scores for depression, anxiety, and stress, while examining the effects of Baseline, Grounded, and Safety prompting conditions on simulation fidelity and bias. Using a quantitative 3×3 factorial design, 720 human–LLM score pairs were analyzed through correlation, Intraclass Correlation Coefficient (ICC), Bland–Altman analysis, error metrics, ANOVA, and chi-square tests. Results showed excellent agreement between human and LLM scores (ICC = .87–.89; r = .87–.89; p < .001) with no significant overall systematic bias (p > .30). However, prompting condition significantly affected simulation error (partial η² = .167; p < .001), with Grounded yielding the lowest and Baseline the highest error. Safety produced significant negative bias for depression (bias = −0.42; p = .025), indicating symptom underestimation. These findings demonstrate high agreement validity while highlighting potential ethical risks of safety-related symptom attenuation.
References
Bland, J. M., & Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. The Lancet, 327(8476), 307–310. https://doi.org/10.1016/S0140-6736(86)90837-8
Bucher, A., Egger, S., Vashkite, I., Wu, W., & Schwabe, G. (2025). "It’s not only attention we need": Systematic review of large language models in mental health care. JMIR Mental Health, 12, e78410. https://doi.org/10.2196/78410
Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6(4), 284–290. https://doi.org/10.1037/1040-3590.6.4.284
Elyoseph, Z., Hadar-Shoval, D., Asraf, K., & Lvovsky, M. (2023). ChatGPT outperforms humans in emotional awareness evaluations. Frontiers in Psychology, 14, 1199058. https://doi.org/10.3389/fpsyg.2023.1199058
Ghasemi, A., & Zahediasl, S. (2012). Normality tests for statistical analysis: A guide for non-statisticians. International Journal of Endocrinology and Metabolism, 10(2), 486–489. https://doi.org/10.5812/ijem.3505
Guo, Z., Lai, A., Thygesen, J. H., Farrington, J., Keen, T., & Li, K. (2024). Large language models for mental health applications: Systematic review. JMIR Mental Health, 11, e57400. https://doi.org/10.2196/57400
Hoang, N. K., Mehri, S., Hsu, T.-A., Sun, Y.-J., Truong, Q. X. N., Doan, K. D., & Hakkani-Tür, D. (2026). PSI-Bench: Towards clinically grounded and interpretable evaluation of depression patient simulators. arXiv preprint arXiv:2604.25840. https://doi.org/10.48550/arXiv.2604.25840
Iftikhar, Z., Xiao, A., Ransom, S., Huang, J., & Suresh, H. (2025). How LLM counselors violate ethical standards in mental health practice: A practitioner-informed framework. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 8(2), 1311–1323. https://doi.org/10.1609/aies.v8i2.36632
Koo, T. K., & Li, M. Y. (2016). A guideline of selecting and reporting intraclass correlation coefficients for reliability research. Journal of Chiropractic Medicine, 15(2), 155–163. https://doi.org/10.1016/j.jcm.2016.02.012
Li, B., Jin, B., Lan, K., Wang, M., & Wu, M. (2026). Synthetic or authentic? Building mental patient simulators from longitudinal evidence. arXiv preprint arXiv:2603.22704. https://doi.org/10.48550/arXiv.2603.22704
Liu, S., Brie, B., Li, W., Biester, L., Lee, A., Pennebaker, J., & Mihalcea, R. (2025). Eeyore: Realistic depression simulation via expert-in-the-loop supervised and preference optimization. Findings of the Association for Computational Linguistics: ACL 2025, 13750–13770. https://doi.org/10.18653/v1/2025.findings-acl.707
Moore, J., Grabb, D., Agnew, W., Klyman, K., Chancellor, S., Ong, D. C., & Haber, N. (2025). Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’25), 599–627. https://doi.org/10.1145/3715275.3732039
Qiu, J., He, Y., Juan, X., Wang, Y., Liu, Y., Yao, Z., Wu, Y., Jiang, X., Yang, L., & Wang, M. (2025). EmoAgent: Assessing and safeguarding human-AI interaction for mental health safety. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, 11741–11756. https://doi.org/10.18653/v1/2025.emnlp-main.594
Rahsepar Meadi, M., Sillekens, T., Metselaar, S., van Balkom, A., Bernstein, J., & Batelaan, N. (2025). Exploring the ethical challenges of conversational AI in mental health care: Scoping review. JMIR Mental Health, 12, e60432. https://doi.org/10.2196/60432
Razali, N. M., & Wah, Y. B. (2011). Power comparisons of Shapiro-Wilk, Kolmogorov-Smirnov, Lilliefors and Anderson-Darling tests. Journal of Statistical Modeling and Analytics, 2(1), 21–33.
Sabour, S., Ng, T., & Huang, M. (2026). PatientHub: A unified framework for patient simulation. arXiv preprint arXiv:2602.11684. https://doi.org/10.48550/arXiv.2602.11684
Sedlakova, J., & Trachsel, M. (2023). Conversational artificial intelligence in psychotherapy: A new therapeutic tool or agent? The American Journal of Bioethics, 23(5), 4–13. https://doi.org/10.1080/15265161.2022.2048739
Shawon, M. T. R., Irbaz, M. S., Elyazori, H. R. A., Resapu, K. R., Lin, Y., Cardenas, V. F., Alemi, F., & Lybarger, K. (2026). Advancing AI trustworthiness through patient simulation: Risk assessment of conversational agents for antidepressant selection. arXiv preprint arXiv:2602.11391. https://doi.org/10.48550/arXiv.2602.11391
Stade, E. C., Stirman, S. W., Ungar, L. H., Boland, C. L., Schwartz, H. A., Yaden, D. B., Sedoc, J., DeRubeis, R. J., Willer, R., & Eichstaedt, J. C. (2024). Large language models could change the future of behavioral healthcare: A proposal for responsible development and evaluation. npj Mental Health Research, 3(1), 12. https://doi.org/10.1038/s44184-024-00056-z
Wang, R., Milani, S., Chiu, J. C., Zhi, J., Eack, S. M., Labrum, T., Murphy, S. M., Jones, N., Hardy, K. V., Shen, H., Fang, F., & Chen, Z. (2024). PATIENT-Ψ: Using large language models to simulate patients for training mental health professionals. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 12772–12797. https://doi.org/10.18653/v1/2024.emnlp-main.711
Wang, X., Pérez, A., Parapar, J., & Crestani, F. (2025). TalkDep: Clinically grounded LLM personas for conversation-centric depression screening. Proceedings of the 34th ACM International Conference on Information and Knowledge Management (CIKM ’25), 6554–6558. https://doi.org/10.1145/3746252.3761617
Xu, Y., Fang, Z., Lin, W., Jiang, Y., Jin, W., Balaji, P., Wang, J., & Xia, T. (2025). Evaluation of large language models on mental health: From knowledge test to illness diagnosis. Frontiers in Psychiatry, 16, 1646974. https://doi.org/10.3389/fpsyt.2025.1646974
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Research and Advancement in Psychological Science and Outcomes (RASO)
This work is licensed under a Creative Commons Attribution 4.0 International License.
