The Path to Self-Evolving Clinical Systems: Scaling Medical Agents from Assistance to Autonomy

Jul 16, 2026

Large language and vision-language models are transforming medical agents from passive, task-specific predictors into autonomous clinical systems that perceive, reason, plan, remember, and act in real environments. This survey reframes the field around deployment realities, defining the tasks, contamination-resistant benchmarks, and interactive training environments required for trust. It formalizes medical agents as sequential decision processes under partial observability, proposes a three-level autonomy taxonomy, and organizes scaling along framework, capability, and environment axes, emphasizing clinical environment scaling and self-evolution.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages50

Reading Time2 h 6 min

Introduction

Preliminaries and Taxonomy

Formal Definition of Medical Agents

Three Level Autonomy Taxonomy

The Scaling Spine: Framework, Capability, and Environment Scaling

Real Environment Clinical Scenarios, Benchmarks, and Training Environments

Real Environment Clinical Scenarios

Benchmarks for Real World Readiness

Evaluation Metrics Beyond Accuracy

Training Environments and Clinical Gyms

Agent Framework Scaling: From Tools to Topologies

Capability Scaling: The Harness-Driven Cognitive Loop

Training Time Scaling: From Imitation to Autonomous Learning

arXiv :2607 . 1 1 175v 1 [ cs .AI] 13 Jul 2026 The Path to Self-Evolving Clinical Systems: Scaling Medical Agents from Assistance to Autonomy Chunzheng Zhu 1 ,∗ , Lei Tian2 ,∗ , Bohan Tan 1 ,∗ , Ziqi Zhou3 ,∗ , Yuxuan Sun4 ,∗ , Yijun Wang 1 , Chengchao Lv 1 , Yilin Wen2 , Yijun He2 , Jinghao Lin2 , Yihang Chen5 , Cheewei Tan6 , Qianshan Wei7 , Lei Zhao8 , Bin Pu 1 , Kenli Li 1 , Yuan Xue9 , Jianxin Lin 1 ,† ∗ Equal Contribution † Corresponding Author 1 Hunan University, 2 ByteDance, 3 Duke University, 4Westlake University, 5 The University of Hong Kong, 6 Nanyang Technological University, 7 Institute of Automation, Chinese Academy of Sciences, 8 University of Macau, 9 The Ohio State University § [https://github.com/zhcz328/Awesome-Medical-Agents](https://github.com/zhcz328/Awesome-Medical-Agents) The growing ability of large language models and vision-language models to jointly interpret and reason over images and text is reshaping medical agents, moving them from passive, task-specific predictors toward autonomous medical agents that perceive, reason, plan, remember, and act within real clinical environments. This survey deliberately departs from the capability-first narrative that dominates concurrent medical agent reviews. We begin from the realities of clinical deployment and ask a fundamental question: what tasks, contamination-resistant benchmarks, and interactive training environments must a medical agent master before it can be trusted in practice? To enable systematic comparison across heterogeneous systems, we formalize such agents as a sequential decision process under partial observability and propose a three level autonomy taxonomy spanning assisted, cooperative, and fully autonomous operation. Around this foundation we organize the entire field along a single scaling spine with three threaded axes, namely framework scaling that covers architecture paradigms and tool orchestration, capability scaling that covers the harness driven loop realizing perception to action, and environment scaling that enriches the tool, data, and interface ecosystem. Within this spine we foreground clinical environment scaling, the synthesis of rich tool and data environments with clinical gyms, as the most immediately actionable yet least explored lever for agents that natively inhabit PACS, EHR, and FHIR ecosystems. Crucially, we position clinical self-evolution, the ability of a medical agent to improve through environment interaction rather than parameter growth alone, as an aspirational frontier rather than a solved capability, and we transfer concrete lessons from general domain self-improving agents, agent gyms, and test-time compute scaling into the imaging setting. Applications across radiology, pathology, ophthalmology, and real hospital workflows are reviewed alongside deployment practice and open risks including hallucination, cascade failures, and fairness. By consolidating over 300 references with particular emphasis on 2025–2026 advances in self-evolution and agentic environments, this survey charts a roadmap toward trustworthy, self-improving medical imaging systems that are genuinely ready for real clinical settings. Keywords: Medical imaging agents, vision-language models, self-evolution, clinical scaling, autonomy taxonomy, agentic benchmarks, trustworthy clinical AI, survey. This survey will be actively maintained and updated together with its companion GitHub repository ([https://github.com/](https://github.com/)[ ](https://github.com/)[zhcz328/Awesome-Medical-Agents](zhcz328/Awesome-Medical-Agents)) . We welcome feedback, corrections, and contributions from the research community. If you identify relevant work that should be included or have suggestions for improving this survey or the repository, please submit an issue or pull request on GitHub, or contact us at [zhuchzh@hnu.edu.cn](zhuchzh@hnu.edu.cn). Contents 1 Introduction 3 2 Preliminaries and Taxonomy 4 2.1 Formal Definition of

FAQs

What core problem does the survey address for building trusted medical agents?

It asks what tasks, contamination-resistant benchmarks, and interactive training environments a medical agent must master before it can be trusted in practice.

How does the survey define autonomy for medical agents?

It formalizes agents as sequential decision processes under partial observability and introduces a three-level autonomy taxonomy covering assisted, cooperative, and fully autonomous operation.

Which three scaling axes does the survey use to organize the field?

The survey organizes scaling along framework scaling (architecture and tool orchestration), capability scaling (a harness-driven loop from perception to action), and environment scaling (tools, data, and interface ecosystems, especially clinical environment scaling).

Jul 16, 2026

Upload by Seraphina