Research

Paper

AI LLM February 26, 2026

MediX-R1: Open Ended Medical Reinforcement Learning

Authors

Sahal Shaji Mullappilly, Mohammed Irfan Kurpath, Omair Mohamed, Mohamed Zidan, Fahad Khan, Salman Khan, Rao Anwer, Hisham Cholakkal

Abstract

We introduce MediX-R1, an open-ended Reinforcement Learning (RL) framework for medical multimodal large language models (MLLMs) that enables clinically grounded, free-form answers beyond multiple-choice formats. MediX-R1 fine-tunes a baseline vision-language backbone with Group Based RL and a composite reward tailored for medical reasoning: an LLM-based accuracy reward that judges semantic correctness with a strict YES/NO decision, a medical embedding-based semantic reward to capture paraphrases and terminology variants, and lightweight format and modality rewards that enforce interpretable reasoning and modality recognition. This multi-signal design provides stable, informative feedback for open-ended outputs where traditional verifiable or MCQ-only rewards fall short. To measure progress, we propose a unified evaluation framework for both text-only and image+text tasks that uses a Reference-based LLM-as-judge in place of brittle string-overlap metrics, capturing semantic correctness, reasoning, and contextual alignment. Despite using only $\sim51$K instruction examples, MediX-R1 achieves excellent results across standard medical LLM (text-only) and VLM (image + text) benchmarks, outperforming strong open-source baselines and delivering particularly large gains on open-ended clinical tasks. Our results demonstrate that open-ended RL with comprehensive reward signals and LLM-based evaluation is a practical path toward reliable medical reasoning in multimodal models. Our trained models, curated datasets and source code are available at https://medix.cvmbzuai.com

Metadata

arXiv ID: 2602.23363
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-02-26
Fetched: 2026-02-27 04:35

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.23363v1</id>\n    <title>MediX-R1: Open Ended Medical Reinforcement Learning</title>\n    <updated>2026-02-26T18:59:46Z</updated>\n    <link href='https://arxiv.org/abs/2602.23363v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.23363v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>We introduce MediX-R1, an open-ended Reinforcement Learning (RL) framework for medical multimodal large language models (MLLMs) that enables clinically grounded, free-form answers beyond multiple-choice formats. MediX-R1 fine-tunes a baseline vision-language backbone with Group Based RL and a composite reward tailored for medical reasoning: an LLM-based accuracy reward that judges semantic correctness with a strict YES/NO decision, a medical embedding-based semantic reward to capture paraphrases and terminology variants, and lightweight format and modality rewards that enforce interpretable reasoning and modality recognition. This multi-signal design provides stable, informative feedback for open-ended outputs where traditional verifiable or MCQ-only rewards fall short. To measure progress, we propose a unified evaluation framework for both text-only and image+text tasks that uses a Reference-based LLM-as-judge in place of brittle string-overlap metrics, capturing semantic correctness, reasoning, and contextual alignment. Despite using only $\\sim51$K instruction examples, MediX-R1 achieves excellent results across standard medical LLM (text-only) and VLM (image + text) benchmarks, outperforming strong open-source baselines and delivering particularly large gains on open-ended clinical tasks. Our results demonstrate that open-ended RL with comprehensive reward signals and LLM-based evaluation is a practical path toward reliable medical reasoning in multimodal models. Our trained models, curated datasets and source code are available at https://medix.cvmbzuai.com</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-02-26T18:59:46Z</published>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Sahal Shaji Mullappilly</name>\n    </author>\n    <author>\n      <name>Mohammed Irfan Kurpath</name>\n    </author>\n    <author>\n      <name>Omair Mohamed</name>\n    </author>\n    <author>\n      <name>Mohamed Zidan</name>\n    </author>\n    <author>\n      <name>Fahad Khan</name>\n    </author>\n    <author>\n      <name>Salman Khan</name>\n    </author>\n    <author>\n      <name>Rao Anwer</name>\n    </author>\n    <author>\n      <name>Hisham Cholakkal</name>\n    </author>\n  </entry>"
}