Med-RewardBench: Benchmarking Reward Models and Judges for Medical Multimodal Large Language Models
This addresses the problem of ensuring accurate and clinically relevant responses from MLLMs in medical applications like disease diagnosis, though it is incremental as it focuses on benchmarking rather than novel model development.
The authors tackled the lack of dedicated benchmarks for evaluating reward models and judges in medical multimodal large language models (MLLMs) by introducing Med-RewardBench, which includes a dataset of 1,026 expert-annotated cases across 13 organ systems and 8 clinical departments, and they evaluated 32 state-of-the-art MLLMs, revealing substantial challenges in aligning outputs with expert judgment.
Multimodal large language models (MLLMs) hold significant potential in medical applications, including disease diagnosis and clinical decision-making. However, these tasks require highly accurate, context-sensitive, and professionally aligned responses, making reliable reward models and judges critical. Despite their importance, medical reward models (MRMs) and judges remain underexplored, with no dedicated benchmarks addressing clinical requirements. Existing benchmarks focus on general MLLM capabilities or evaluate models as solvers, neglecting essential evaluation dimensions like diagnostic accuracy and clinical relevance. To address this, we introduce Med-RewardBench, the first benchmark specifically designed to evaluate MRMs and judges in medical scenarios. Med-RewardBench features a multimodal dataset spanning 13 organ systems and 8 clinical departments, with 1,026 expert-annotated cases. A rigorous three-step process ensures high-quality evaluation data across six clinically critical dimensions. We evaluate 32 state-of-the-art MLLMs, including open-source, proprietary, and medical-specific models, revealing substantial challenges in aligning outputs with expert judgment. Additionally, we develop baseline models that demonstrate substantial performance improvements through fine-tuning.