SkeMex is a training-free self-evolution framework that allows medical agents to accumulate, organize, and reuse valuable reasoning skills from experience, leading to more effective and generalizable clinical decision-making.</p>\n","updatedAt":"2026-06-09T05:50:04.357Z","author":{"_id":"67547707f168984215451697","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67547707f168984215451697/7Hp8Neb8_jkW4m0LXXXiM.jpeg","fullname":"Manglu","name":"manglu3935","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":2,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9093718528747559},"editors":["manglu3935"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/67547707f168984215451697/7Hp8Neb8_jkW4m0LXXXiM.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.09365","authors":[{"_id":"6a27a91f6dde1c5ef75bd152","name":"Haoran Sun","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd153","name":"Wenjie Li","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd154","name":"Yujie Zhang","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd155","name":"Zekai Lin","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd156","name":"Fanrui Zhang","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd157","name":"Kaitao Chen","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd158","name":"Xingqi He","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd159","name":"Yichen Li","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd15a","name":"Mianxin Liu","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd15b","name":"Lei Liu","hidden":false},{"_id":"6a27a91f6dde1c5ef75bd15c","name":"Yankai Jiang","hidden":false}],"publishedAt":"2026-06-08T11:37:01.000Z","submittedOnDailyAt":"2026-06-09T00:00:00.000Z","title":"Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory","submittedOnDailyBy":{"_id":"67547707f168984215451697","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67547707f168984215451697/7Hp8Neb8_jkW4m0LXXXiM.jpeg","isPro":false,"fullname":"Manglu","user":"manglu3935","type":"user","name":"manglu3935"},"summary":"Medical agent systems are increasingly expected to support interactive clinical decision making rather than only static question answering. In such settings, effective agents must reuse prior experience across evolving cases, yet existing memory mechanisms often retain raw historical traces that are redundant, noisy, and difficult to govern. More importantly, they rarely distinguish which memories are truly useful for future reasoning. This limits their ability to accumulate compact and reliable experience for long-horizon clinical reasoning. To close this gap, we propose SkeMex, a post-deployment self-evolution framework that improves medical agents through a skill-based memory without updating model weights. SkeMex distills informative interaction trajectories into structured skills that encode reusable procedural knowledge, and organizes them into a multi-branch repository spanning general, task-specific, and action-level experience. To determine which memories should be reused and retained, SkeMex estimates context-dependent utility from environment feedback and uses it to guide value-aware retrieval and repository governance. A closed-loop ``Read--Write--Assess--Govern\" lifecycle further supports continual evolution by writing new skills, updating utilities, promoting useful memories, and removing harmful entries. Experiments across diverse clinical tasks show that SkeMex consistently outperforms representative memory-based agents in both offline and online settings. It also generalizes across model backbones and supports transferable skill memory. All data and code will be released publicly.","upvotes":1,"discussionId":"6a27a9206dde1c5ef75bd15d","ai_summary":"SkeMex is a self-evolving framework that enhances medical agents through structured skill memory, improving long-term clinical reasoning by distinguishing useful experiences and governing memory retention based on contextual utility.","ai_keywords":["medical agent systems","interactive clinical decision making","memory mechanisms","post-deployment self-evolution","skill-based memory","interaction trajectories","procedural knowledge","multi-branch repository","context-dependent utility","value-aware retrieval","repository governance","closed-loop lifecycle","Read--Write--Assess--Govern","continual evolution","offline and online settings","transferable skill memory"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct"},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"67547707f168984215451697","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67547707f168984215451697/7Hp8Neb8_jkW4m0LXXXiM.jpeg","isPro":false,"fullname":"Manglu","user":"manglu3935","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.09365.md"}">
Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory
Published on Jun 8
· Submitted by Manglu on Jun 9 Authors: ,
,
,
,
,
,
,
,
,
,
Abstract
SkeMex is a self-evolving framework that enhances medical agents through structured skill memory, improving long-term clinical reasoning by distinguishing useful experiences and governing memory retention based on contextual utility.
Medical agent systems are increasingly expected to support interactive clinical decision making rather than only static question answering. In such settings, effective agents must reuse prior experience across evolving cases, yet existing memory mechanisms often retain raw historical traces that are redundant, noisy, and difficult to govern. More importantly, they rarely distinguish which memories are truly useful for future reasoning. This limits their ability to accumulate compact and reliable experience for long-horizon clinical reasoning. To close this gap, we propose SkeMex, a post-deployment self-evolution framework that improves medical agents through a skill-based memory without updating model weights. SkeMex distills informative interaction trajectories into structured skills that encode reusable procedural knowledge, and organizes them into a multi-branch repository spanning general, task-specific, and action-level experience. To determine which memories should be reused and retained, SkeMex estimates context-dependent utility from environment feedback and uses it to guide value-aware retrieval and repository governance. A closed-loop ``Read--Write--Assess--Govern" lifecycle further supports continual evolution by writing new skills, updating utilities, promoting useful memories, and removing harmful entries. Experiments across diverse clinical tasks show that SkeMex consistently outperforms representative memory-based agents in both offline and online settings. It also generalizes across model backbones and supports transferable skill memory. All data and code will be released publicly.
Community
SkeMex is a training-free self-evolution framework that allows medical agents to accumulate, organize, and reuse valuable reasoning skills from experience, leading to more effective and generalizable clinical decision-making.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.09365 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.09365 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.09365 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.