Reinforcement learning improves LLM accuracy and reasoning in disease classification from radiology reports.
Authors
Wei Y,Lin Y,Flanders A,Shih G,Peng Y
Affiliations (5)
Affiliations (5)
- Department of Population Health Sciences, Weill Cornell Medicine, New York, NY, USA.
- Department of Radiology, Weill Cornell Medicine, New York, NY, USA.
- Department of Radiology, Thomas Jefferson University, Philadelphia, PA, USA.
- Department of Population Health Sciences, Weill Cornell Medicine, New York, NY, USA. [email protected].
- Department of Radiology, Weill Cornell Medicine, New York, NY, USA. [email protected].
Abstract
Accurate disease classification from radiology reports is essential for many applications. While supervised fine-tuning (SFT) of lightweight LLMs improves accuracy, it can degrade reasoning. We propose a two-stage approach: SFT on disease labels followed by Group Relative Policy Optimization (GRPO) to refine predictions by optimizing accuracy and format without reasoning supervision. Across three radiologist-annotated datasets, SFT outperformed baselines and GRPO further improved classification and enhanced reasoning recall and comprehensiveness.
Topics
Journal Article