---
title: "Training language models in simulation for psychiatric intake"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.40476v1"
permalink: "https://feed.astralor.com/items/itm_4bcb1e217d72eba189ebfa15"
published: "2026-10-29T17:10:32.000Z"
score: 58
channel: "ai"
tags: ["AI Agent","安全","Training","Here","RL-trained","Across","Delphi-consensus","Simulation"]
---

# Training language models in simulation for psychiatric intake

## 速览

In psychiatric practice, the clinical intake interview remains a primary diagnostic instrument, requiring the clinician to establish rapport while disentangling complex, overlapping presentations. Here, we show that mul…

## 关键点

- In psychiatric practice, the clinical intake interview remains a primary diagnostic instrument, requiring the clinician to establish rapport while disentangling complex, overlapping presentations.
- Here, we show that multi-turn reinforcement learning (RL) in simulation environments improves language model performance on structured text-based psychiatric intake, spanning history-taking, differential diagnosis, and …
- In a randomized, double-blind crossover study ($N = 60$ matched scenario pairs, 120 encounters) comparing the RL-trained agent against its identically prompted base model, 32 board-certified psychiatrists role-played st…

## 全文

In psychiatric practice, the clinical intake interview remains a primary diagnostic instrument, requiring the clinician to establish rapport while disentangling complex, overlapping presentations. Here, we show that multi-turn reinforcement learning (RL) in simulation environments improves language model performance on structured text-based psychiatric intake, spanning history-taking, differential diagnosis, and initial management planning. In a randomized, double-blind crossover study ($N = 60$ matched scenario pairs, 120 encounters) comparing the RL-trained agent against its identically prompted base model, 32 board-certified psychiatrists role-played standardized patients while a separate panel of 15 independent, board-certified psychiatrists scored the de-identified transcripts. The trained agent outperformed the base model in 76.7% of matched scenario pairs ($p < 0.0001$), with gains on an adapted clinical-examination rubric ($d_z = 0.77$, $p < 0.0001$), patient-centered communication (82.0% vs. 65.2%, $p < 0.001$), and differential diagnosis appropriateness (68.3% vs. 33.3%, $p < 0.001$). Across 127 psychiatrist-validated simulated intake scenarios, the trained agent nearly doubled its base model's clinical rubric score (0.706 vs. 0.379; composite reward 0.564 vs. 0.294, $p < 0.0001$). To systematically probe conversational safety, we established two Delphi-consensus taxonomies, of clinical safety risks and relational competencies, from which 256 challenging adversarial presentations were derived; we find that training in simulation environments more than halved under-triage of acute safety risks (12.9% vs. 30.5%, $p < 0.0001$). Simulation environments thus provide a controlled testbed for training and stress-testing psychiatric conversational agents, with gains that transferred to psychiatrist-enacted encounters, though residual gaps require safeguards and clinician oversight.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_4bcb1e217d72eba189ebfa15