Reproduce “Scaffolding Empathy: Training Counselors with Simulated Patients and Utterance-level Performance Visualizations” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ian Steenstra, Farnaz Nouraei, Timothy W. Bickmore
Northeastern University researchers developed SimPatient, an LLM-powered multi-agent system for Motivational Interviewing training that provides counselors with utterance-level feedback and visualizations of dynamic patient cognitive states. The system significantly increased trainee self-efficacy, achieved an "Excellent" System Usability Scale score of 88.1, and demonstrated high utility for evaluating counseling skills.
Key insights
- LLMs can effectively simulate patient personas for specific characteristics like gender and occupation, though fidelity for more complex traits such as age or personality type remains a challenge.
- Visualizing dynamic patient cognitive states in real-time allows counselors to better understand the impact of their interventions and adapt their approach, even if some factor fluctuations (e.g., Reward) require further refinement for realism.
- Detailed, transparent, and individualized feedback at the utterance level, coupled with annotated transcripts and competency thresholds, significantly enhances trainee self-reflection and motivates targeted skill improvement in counseling.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.