Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.12210 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915940195631104 |
|---|---|
| author | Zhang, Weikang Zhu, Zimo Yang, Zhichuan Huang, Chen Lei, Wenqiang Ng, See-Kiong |
| author_facet | Zhang, Weikang Zhu, Zimo Yang, Zhichuan Huang, Chen Lei, Wenqiang Ng, See-Kiong |
| contents | Simulating Standardized Patients with cognitive impairment offers a scalable and ethical solution for clinical training. However, existing methods rely on discrete prompt engineering and fail to capture the heterogeneity of deficits across varying domains and severity levels. To address this limitation, we propose StsPatient for the fine-grained simulation of cognitively impaired patients. We innovatively capture domain-specific features by extracting steering vectors from contrastive pairs of instructions and responses. Furthermore, we introduce a Stochastic Token Modulation (STM) mechanism to regulate the intervention probability. STM enables precise control over impairment severity while mitigating the instability of conventional vector methods. Comprehensive experiments demonstrate that StsPatient significantly outperforms baselines in both clinical authenticity and severity controllability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_12210 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Beyond Prompt: Fine-grained Simulation of Cognitively Impaired Standardized Patients via Stochastic Steering Zhang, Weikang Zhu, Zimo Yang, Zhichuan Huang, Chen Lei, Wenqiang Ng, See-Kiong Artificial Intelligence Computation and Language Simulating Standardized Patients with cognitive impairment offers a scalable and ethical solution for clinical training. However, existing methods rely on discrete prompt engineering and fail to capture the heterogeneity of deficits across varying domains and severity levels. To address this limitation, we propose StsPatient for the fine-grained simulation of cognitively impaired patients. We innovatively capture domain-specific features by extracting steering vectors from contrastive pairs of instructions and responses. Furthermore, we introduce a Stochastic Token Modulation (STM) mechanism to regulate the intervention probability. STM enables precise control over impairment severity while mitigating the instability of conventional vector methods. Comprehensive experiments demonstrate that StsPatient significantly outperforms baselines in both clinical authenticity and severity controllability. |
| title | Beyond Prompt: Fine-grained Simulation of Cognitively Impaired Standardized Patients via Stochastic Steering |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2604.12210 |