analiza aiqbench
This study presents ClinMPO, a reinforcement learning framework that successfully aligns small language models with expert psychiatric reasoning strategies without relying on massive datasets. By incorporating psychiatrist-defined guidelines and clinical evidence into the reward model, the approach enables smaller architectures to surpass human medical student baselines in diagnostic accuracy and rationale quality. The work demonstrates a viable path for deploying specialized AI tools in resource-constrained healthcare settings where large models are impractical.
An evidence-guided reinforcement learning method to improve psychiatric reasoning in small language models
arXiv:2602.06449v2 Announce Type: replace Abstract: Privacy and computational constraints limit the use of large language models in psychiatry, while adapting small language models (SLMs) often requires substantial data and expert annotation. We developed ClinMPO, an evidence-guided reinforcement-learning framework guided by the psychiatrist-defined Clinical Psychiatry Thinking Strategy (CPTS). ClinMPO uses ClinRM, a reward model trained on 18,569 question--answer pairs from 4,474 psychiatry articles. We evaluated four Qwen3 sizes on 1,737 model-screened questions. ClinMPO outperformed Base, supervised fine-tuning and standard group relative policy optimization across scales. From responses by 300 senior pre-licensure medical students, we established the human baseline, a medical-student reference. The 4B model approached this baseline, whereas the 8B model surpassed it and ranked first among 31 models and post-training variants. ClinMPO improved performance across two complementary schemes covering ICD-11 diagnostic categories and psychiatric practice competencies. Blinded assessment by three clinicians showed improved rationale quality across CPTS criteria. These findings highlight how existing clinical evidence and specialist knowledge can be incorporated into the development of medical AI systems through evidence-guided learning.
Authors:Xinxin Lin, Guangxin Dai, Yi Zhong, Xiang Li, Xue Xiao, Jian Liu, Yixin Zhang, Lingming Hu, Zhengdong Wu, Yongbo Zheng, Runchuan Zhu, Ming Zhao, Huizi Yu, Yi Zhang, Fangting Lu, Shuo Wu, Jun Zhao, Ping Yin, Joey W. Y. Chan, Ngan Yin Chan, Yumei Wang, Lejin Yang, Yanqiu Xing, Sijing Chen, Yun Kwok Wing, Lin Lu, Xin Ma, Lizhou Fan
Abstract:Privacy and computational constraints limit the use of large language models in psychiatry, while adapting small language models (SLMs) often requires substantial data and expert annotation. We developed ClinMPO, an evidence-guided reinforcement-learning framework guided by the psychiatrist-defined Clinical Psychiatry Thinking Strategy (CPTS). ClinMPO uses ClinRM, a reward model trained on 18,569 question--answer pairs from 4,474 psychiatry articles. We evaluated four Qwen3 sizes on 1,737 model-screened questions. ClinMPO outperformed Base, supervised fine-tuning and standard group relative policy optimization across scales. From responses by 300 senior pre-licensure medical students, we established the human baseline, a medical-student reference. The 4B model approached this baseline, whereas the 8B model surpassed it and ranked first among 31 models and post-training variants. ClinMPO improved performance across two complementary schemes covering ICD-11 diagnostic categories and psychiatric practice competencies. Blinded assessment by three clinicians showed improved rationale quality across CPTS criteria. These findings highlight how existing clinical evidence and specialist knowledge can be incorporated into the development of medical AI systems through evidence-guided learning.
Computation and Language (cs.CL)
arXiv:2602.06449 [cs.CL]
(or arXiv:2602.06449v2 [cs.CL] for this version)
Text integral sub licența deschisă a publicației · 238 cuvinte
Subiecte conexe
Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk
Latent Space · 8 oct. 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · 18 sept. 2026
ReST-RL: Reinforcing LLM Reasoning through Unified Self-Training and Value-Guided Search
arXiv · 9 sept. 2026
Spatial Memory Agent: Experience-Grounded Procedure Memory for Spatial Intelligence
arXiv · 29 sept. 2026