Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2511.14783 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912978215895040 |
|---|---|
| author | Zhang, Bingquan Liu, Xiaoxiao Wang, Yuchi Zhou, Lei Xie, Qianqian Wang, Benyou |
| author_facet | Zhang, Bingquan Liu, Xiaoxiao Wang, Yuchi Zhou, Lei Xie, Qianqian Wang, Benyou |
| contents | Standardized patients (SPs) are indispensable for clinical skills training but remain expensive and difficult to scale. Although large language model (LLM)-based virtual standardized patients (VSPs) have been proposed as an alternative, their behavior remains unstable and lacks rigorous comparison with human standardized patients. We propose EasyMED, a multi-agent VSP framework that separates case-grounded information disclosure from response generation to support stable, inquiry-conditioned patient behavior. We also introduce SPBench, a human-grounded benchmark with eight expert-defined criteria for interaction-level evaluation. Experiments show that EasyMED more closely matches human SP behavior than existing VSPs, particularly in case consistency and controlled disclosure. A four-week controlled study further demonstrates learning outcomes comparable to human SP training, with stronger early gains for novice learners and improved flexibility, psychological safety, and cost efficiency. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_14783 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Human or LLM as Standardized Patients? A Comparative Study for Medical Education Zhang, Bingquan Liu, Xiaoxiao Wang, Yuchi Zhou, Lei Xie, Qianqian Wang, Benyou Computation and Language Computers and Society Standardized patients (SPs) are indispensable for clinical skills training but remain expensive and difficult to scale. Although large language model (LLM)-based virtual standardized patients (VSPs) have been proposed as an alternative, their behavior remains unstable and lacks rigorous comparison with human standardized patients. We propose EasyMED, a multi-agent VSP framework that separates case-grounded information disclosure from response generation to support stable, inquiry-conditioned patient behavior. We also introduce SPBench, a human-grounded benchmark with eight expert-defined criteria for interaction-level evaluation. Experiments show that EasyMED more closely matches human SP behavior than existing VSPs, particularly in case consistency and controlled disclosure. A four-week controlled study further demonstrates learning outcomes comparable to human SP training, with stronger early gains for novice learners and improved flexibility, psychological safety, and cost efficiency. |
| title | Human or LLM as Standardized Patients? A Comparative Study for Medical Education |
| topic | Computation and Language Computers and Society |
| url | https://arxiv.org/abs/2511.14783 |