Training language models in simulation for psychiatric intake

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.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-10-29
DOI
https://doi.org/10.48550/arxiv.2610.40476
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Training language models in simulation for psychiatric intake

Aleksandra Faust, Ellen E. Fitzsimmons‐Craft, Jared Moore, Elizabeth Cameron Stade et al.
arXiv (Cornell University)
Artificial Intelligence in Healthcare and Education
preprint

Training language models in simulation for psychiatric intake

Aleksandra Faust, Ellen E. Fitzsimmons‐Craft, Jared Moore, Elizabeth Cameron Stade, Desmond C. Ong, Kameron Collin Black, Théo Moutakanni, Alvin Rajkomar, Paul Jhun, Mike Schaekermann, John Blake Torous, David Wu, Valentin Liévin, Caroline Astrid Figueroa, Po-Hsuan Cameron Chen, Anil K. Palepu, A. Bijamov, Ulysse Marteau-Ferey, Sarah Montgomery‐Taylor, Joelle K. Barral, Eric C. Lin, Neguine Rezaii, Logan Pierce, S. Sara Mahdavi, Joseph Breda, Katharina Schultebraucks, Mark S. Chen, Samuel Schmidgall, johannes Christopher Eichstaedt, Austin Schoeffler, Eun Kyoung Hong, Shashank Ravi, Johannes Thrul, Min Sun, Isaac R. Galatzer‐Levy, Tim Strother, Daniel McDuff, David Racz, Bhavna Daryani, Ernest Lim, Shirley Wang, Torrey Creed, Yun Liu, Thomas H. McCoy, Justin Chen, Jeff Jopling, John Lyon Havlik, Zhihan Jiang, Xuhai "Orson" Xu, Lin Yang, Yubin Kim, Christina Chen, Andrew Marshall, Tao Tu, Akshay Goel, Kate H. Bentley
preprint en

Abstract

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.

arXiv (Cornell University)
Artificial Intelligence in Healthcare and Education
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon