G-CARL: Grounded Checklist-Aligned Reward Learning for Patient-Oriented Medical Report Interpretation

Shiao Xie; Siyu Chen; Jianwei Lv; Bo Yuan; Yujin Wang; Xiandong Li

cs.CLAug 20, 2026

G-CARL: Grounded Checklist-Aligned Reward Learning for Patient-Oriented Medical Report Interpretation

Authors: Shiao Xie, Siyu Chen, Jianwei Lv, Bo Yuan, Yujin Wang, Xiandong Li

Organizations: Baidu Inc., China

Abstract

Personalized interpretation of medical reports has emerged as an increasingly important need among patients. Addressing this need requires both evidence-grounded medical factuality and context-dependent patient communication, yet existing medical vision-language tasks do not adequately capture these dual requirements. To bridge this gap, we introduce Patient-oriented Medical Report Interpretation (PMRI), a novel open-ended multimodal generation task that requires models to explain medical reports in accurate and accessible language based on a user's query and dialogue history. These two objectives differ fundamentally in their verifiability, yet remain tightly coupled, making them difficult to optimize jointly under conventional supervised fine-tuning and holistic reinforcement learning paradigms. To address this challenge, we propose G-CARL, a grounded, checklist-aligned reinforcement learning framework that combines multi-source retrieval for atomic claim verification with context-aware, instance-specific weighted checklists for response coverage, providing structured supervision for factuality, user-demand satisfaction, and expression quality without constraining response diversity. We further construct MMedReport, a real-world PMRI benchmark, along with a clinician-designed three-dimensional evaluation protocol. Extensive experiments demonstrate that G-CARL consistently outperforms existing post-training baselines in overall quality, claim-level precision, and checklist recall. Pairwise preference evaluation by clinicians further confirms that G-CARL produces interpretations that are more accurate and better aligned with patient needs.

Figures & tables

Figure 1: Comparison of PMRI ( Task C ) with medical VQA ( Task A ) and conventional report generation ( Task B ). Unlike the short, deterministic outputs of Tasks A and B, PMRI produces long-form, patient-facing explanations grounded in multimodal reports and extended patient–doctor dialogue.

Figure 2: Overview of G-CARL. Given report images, user queries, and dialogue history, G-CARL samples $G$ responses from the old policy and optimizes them with a multi-objective reward function consisting of retrieval-grounded claim reward $R_{\mathrm{fact}}$ , case-specific checklist reward $R_{\mathrm{check}}$ , and structured reasoning format reward $R_{\mathrm{format}}$ .

Method	Subjective Metrics				Objective Metrics
Method	Overall	Accuracy	Satisfaction	Expression	Precision (%)	Recall (%)
General LVLMs
GPT-4o ( Hurst et al. 2024 )	1.588 ±0.007	0.974 ±0.007	0.358 ±0.001	0.256 ±0.001	95.26	42.63
ERNIE 4.5 VL ( Baidu ERNIE Team 2025 )	1.709 ±0.010	1.088 ±0.012	0.376 ±0.002	0.245 ±0.000	95.27	53.74
GLM-4.6V ( Team and others 2026 )	1.819 ±0.014	1.159 ±0.013	0.395 ±0.002	0.265 ±0.001	95.90	59.51
Step-3.7-Flash ( StepFun AI 2026 )	1.842 ±0.011	1.175 ±0.012	0.412 ±0.002	0.255 ±0.001	95.96	72.11

Table 1: Main results on our medical report interpretation benchmark. Numbers are mean over 3 seeds with standard deviation.

Method	QA Accuracy (%)		Open-ended Generation
Method	Train	Val	Professionalism
Base	74.85	71.42	3.58
SFT	75.00	71.43	3.48
MLLM-as-a-Judge	74.96	70.36	3.51
G-CARL	75.48	72.05	3.61

Table 2: Evaluation results on the CMB dataset.

Reward Combination
	Overall	Acc	Sat	Exp	Pre	Rec
Baseline	1.626	0.980	0.388	0.258	93.46	60.68
+ $R_{\text{check}}$ + $R_{\text{format}}$	1.720	1.046	0.406	0.268	95.36	64.10
+ $R_{\text{fact}}$ + $R_{\text{format}}$	1.748	1.074	0.404	0.270	95.84	62.81
+ $R_{\text{fact}}$ + $R_{\text{check}}$	1.810	1.135	0.411	0.264	96.03	70.09
+ $R_{\text{fact}}$ + $R_{\text{check}}$ + $R_{\text{format}}$	1.829	1.141	0.411	0.277	96.62	72.18

Table 3: Ablation study of reward designs in G-CARL.

Figure 3: Qualitative comparison of interpretation results across different models. Two representative real-world cases (Case A and Case B) are presented, where Case B includes both report images and dialogue history as input.

Figure 4: Human preference evaluation by clinicians and patients. "Ours/B", "Ours/S", and "Ours/M" denote pairwise comparisons between G-CARL and the base model, the SFT model, and the MLLM-as-a-Judge-trained model.

Method	Overall	Acc	Sat	Exp
DPO ( Rafailov et al. 2023 )	1.032	0.438	0.332	0.262
GRPO-based methods
+ PROMETHEUS ( Kim et al. 2024 )	1.132	0.530	0.347	0.255
+ RAR ( Gunjal et al. 2025 )	1.683	1.016	0.398	0.269
+ MedRepBench ( Shang et al. 2025 )	1.722	1.048	0.406	0.268
+ CapRL ( Xing et al. 2025 )	1.729	1.050	0.410	0.269

Table 4: Comparison with existing RL methods.

Medical VLMs

Medical vision-language models connect clinical images with text for diagnosis, reporting, question answering, and image grounding.

See all 475 papers

MedRepBench: A Comprehensive Benchmark for Medical Report Interpretation
Aug 21, 2025Fangxin Shang, Yuan Xia, Dalu Yang +2VLM Evaluation Healthcare
Baidu Inc., China
MediX-R1: Open Ended Medical Reinforcement Learning
Feb 26, 2026Sahal Shaji Mullappilly, Mohammed Irfan Kurpath, Omair Mohamed +5RL for VLMs Medical VLMs
Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) · Jubilee Mission Medical College and Research Institute · JJM Medical College
MedPRMBench: A Fine-grained Benchmark for Process Reward Models in Medical Reasoning
Apr 19, 2026Xiang Zheng, Lingyan Wu, Weiqi Zhai +4Healthcare Process Reward Models
Alibaba Group, China

Aug 21, 2025J/K move · Enter open · S save

MedRepBench: A Comprehensive Benchmark for Medical Report Interpretation

Fangxin Shang, Yuan Xia, Dalu Yang, Yahui Wang+1

Baidu Inc., China

VLM Evaluation Healthcare Clinical Information Extraction Document Image Understanding

Open paper AskPDF arXiv

Medical report understanding from real-world document images is essential for generating patient-facing explanations and enabling structured information exchange in clinical systems. Existing VLMs and LLMs have shown strong performance on document understanding, but structured understanding of medical reports remains insufficiently benchmarked. Therefore, we introduce MedRepBench, a benchmark with 1,925 de-identified Chinese medical report images spanning diverse departments, patient demographics, and acquisition formats. In MedRepBench, we mainly focus on report-grounded interpretation rather than evaluating diagnostic reasoning, treatment recommendation, or the integration of patient history. The interpretation is defined as structured extraction of report fields (e.g., item, value, unit, reference range, abnormal flag) plus a patient-facing explanation grounded strictly in the report content. The benchmark primarily evaluates end-to-end VLMs, and also includes a controlled text-only setting (high-quality OCR + LLM) to approximate an upper bound when character recognition errors are minimized. Our evaluation framework provides two complementary protocols: (1) an objective protocol measuring field-level recall of structured items, and (2) an automated subjective protocol that uses an LLM-based judge to score factuality, interpretability, and reasoning quality under a fixed prompt. Using the objective metric as a reward signal, we also provide a lightweight GRPO-based alignment baseline for a mid-sized VLM, which improves field-level recall by up to 6%. Finally, we analyze practical limitations of OCR+LLM pipelines, including layout-related errors and additional system latency, showing the need for robust end-to-end vision-based medical report understanding. The dataset and evaluation resources are publicly available on https://huggingface.co/datasets/MedRepBench/MedRepBench.

Abstract

Medical VLMs

Explore similar work

MedRepBench: A Comprehensive Benchmark for Medical Report Interpretation

MediX-R1: Open Ended Medical Reinforcement Learning

MedPRMBench: A Fine-grained Benchmark for Process Reward Models in Medical Reasoning