cs.AINov 20, 2025

Knowledge and Reasoning Augmented Learning for Language Model Assisted Antimicrobial Decision Support

Authors: Zhe Li, Yehan Qiu, Yuntian Huang, Xin Ding, Yujie Chen, Jing Sun, Min Yan, Lei Chen, 

Organizations: Information Center, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Beijing, 100730, China · Department of Critical Care Medicine, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing, 100730, China · Medical Intensive Care Unit, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing, 100730, China

Abstract

Background: Antimicrobial decision support requires both clinical knowledge and patient-specific reasoning. We developed Knowledge and Reasoning Augmented Learning (KRAL) to improve language model performance on antimicrobial decision-making tasks. Methods: In this retrospective technical study, a 27-billion-parameter student model underwent guideline-grounded supervised fine-tuning followed by Multi-turn Agentic GRPO using clinical reasoning trajectories. Six configurations were evaluated on 123 clinician-validated clinical antimicrobial treatment cases from Peking Union Medical College Hospital (PUMCH) and 100 JAMA Clinical Challenge cases. A language model judge scored responses, and paired tests compared configurations. Component ablations assessed benchmark performance and training stability. Results: On the PUMCH benchmark, KRAL achieved the highest composite score (0.895 on a 0-1 scale), outperforming all five controls (all P<0.001). On JAMA, KRAL scored 0.600 and exceeded the base model with and without retrieval (P = 0.0048 and 0.0499, respectively). Reinforcement learning ablation experiments demonstrate that the reward smoothing can significantly improve the performance and training stability of vanilla GRPO. Conclusions: KRAL improved performance on a single-center clinical antimicrobial benchmark and publicly available JAMA clinical challenge cases by combining guideline-grounded learning with Multi-turn Agentic GRPO. Further external clinical validation is needed to evaluate the practical utility and effectiveness of KRAL.

Explore similar work

CardsList
  1. MedGuideX: Internalizing Decision Logic from Executable Guidelines into Large Language Models for Clinical Reasoning

    May 26, 2026Yuhao Shen, Lang Cao, Simo Du +4HealthcareClinical Decision Support

  2. DeepImagine: Learning Biomedical Reasoning via Successive Counterfactual Imagining

    Apr 24, 2026Youze Zheng, Jianyou Wang, Yuhan Chen +8LLM Fine-TuningCounterfactual Reasoning in Language Models