[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-327635-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-327635-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","reinforcement-learning-for-evidence-seeking-diagnostic-reasoning-with-large-language-models","Reinforcement Learning for Evidence-Seeking Diagnostic Reasoning with Large Language Models","","Recent reasoning-centric Large Language Models (LLMs) advance in inference, yet largely assume complete information, unlike clinical intelligence that unfolds as an iterative investigation requiring strategic evidence gathering. The work formalizes medical diagnosis as an Iterative Evidence-Seeking Task and applies Reinforcement Learning with Verifiable Rewards (RLVR) to induce intrinsic reasoning in a closed loop. A Retrieval-Augmented Generation-based Examination Simulator (RAGES) serves as a verifiable clinical oracle to supply realistic follow-up evidence, improving diagnostic precision and examination consistency across datasets.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/reinforcement-learning-for-evidence-seeking-diagnostic-reasoning-with-large-language-models/327635/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/reinforcement-learning-for-evidence-seeking-diagnostic-reasoning-with-large-language-models/327635.png","ImageObject",300,407,{"name":42,"@type":43},"Aria","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-04","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",23,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"How does the paper address the mismatch between LLM inference and real clinical workflows?","Question",{"text":63,"@type":64},"It models medical diagnosis as an iterative evidence-seeking task rather than a one-shot inference problem, reflecting the active investigative nature of clinical practice.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"What is RLVR used for in this framework?",{"text":68,"@type":64},"Reinforcement Learning with Verifiable Rewards (RLVR) is used to elicit intrinsic reasoning in a closed-loop environment, guided by reward signals that enforce diagnostic precision and consistency.",{"name":70,"@type":61,"acceptedAnswer":71},"What role does RAGES play in the proposed system?",{"text":72,"@type":64},"RAGES is a retrieval-augmented examination simulator that acts as a high-fidelity clinical oracle, generating realistic follow-up evidence to support an autonomous reasoning loop.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},327635,1790166144,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":140,"language":141,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":12,"update_tm":145,"read_time":146,"preview_page_count":4},2336464648322,"https://ap-avatar.wpscdn.com/avatar/2200025388227c56fec?_k=1778556882303663488","arXiv :2607 .02983v 1 [ cs .AI] 3 Jul 2026  \nReinforcement Learning for Evidence-Seeking Diagnostic Reasoning with Large Language Models  \nShengyi Hua1, Kangzhe Hu1, Conghui He3, Xiaofan Zhang1,2,∗, Shaoting Zhang1,∗ 1 Qing Yuan Research Institute, Shanghai Jiao Tong University  \n2 Shanghai Innovation Institute  \n3 Shanghai AI Laboratoty  \n∗ Corresponding authors: [xiaofan.zhang@sjtu.edu.cn](xiaofan.zhang@sjtu.edu.cn)  \nAbstract  \nRecent reasoning-centric Large Language Models (LLMs) have made significant strides, yet they predominantly operate on a passive-inference pattern that assumes complete information. In contrast, real-world clinical intelligence is inherently an iterative investigative process requiring strategic evidence acquisition. To bridge this gap, we formalize medical diagnosis as an Iterative Evidence-Seeking Task. We leverage Reinforcement Learning with Verifiable Rewards (RLVR) to elicit intrinsic reasoning within a closed-loop environment, guided by a novel suite of rewards that enforce diagnostic precision and examination consistency. To facilitate this, we introduce the Retrieval-Augmented Generation-based Examination Simulator (RAGES), a high-fidelity clinical oracle that provides realistic, knowledge-grounded follow-up evidence. Empirical results across diverse datasets demonstrate that our framework enables LLMs to transition from passive respondersto autonomous assistants. Notably, our model demonstrates comparable performance to larger and reasoning-enhanced baselines, while RAGES proves superior to vanilla LLMs in generating biologically plausible clinical feedback.  \n1 Introduction  \nThe rise of reasoning-centric models like OpenAI o1 (Jaech et al., 2024) and DeepSeekR1 (DeepSeek-AI, 2025) has shifted the LLM landscape toward sophisticated chain-ofthought capabilities. However, a gap remains between static benchmarks and the dynamic nature of real-world intelligence. True intelligence requires more than one-shot inference. It demands the autonomous capacity to actively seek and integrate missing information to resolve ambiguity. This necessity is most acute in the medical domain, where evidence-based diagnosis serves as the gold standard for clinical safety (Emmi et al., 2023) . In clinical practice, diagnosis is inherently an active investigative process: clinicians start with incomplete profiles and must strategically order examinations to refine their hypotheses. Pathological diagnosis, in particular, represents this multi-turn diagnostic trajectory, requiring a seamless loop between hypothesis generation and evidence acquisition.  \nFor LLMs, mastering this process is non-trivial, as it demands both deep medical knowledge and strategic reasoning to prioritize examinations under uncertainty. Current medical LLM research, however, is largely confined to a passive-inference pattern, focusing on single-turn, multiple-choice benchmarks with complete information (Qiu et al., 2025b; Zhang et al., 2025) . These settings fail to capture the challenges of partial observability. Furthermore, the inherent subjectivity of differential diagnosis complicates the objective evaluation.  \nTo bridge these gaps, we formalize medical diagnosis as an Iterative Evidence-Seeking Task. We employ Reinforcement Learning with Verifiable Rewards (RLVR) (DeepSeek-AI, 2025; Mroueh, 2025) to elicit intrinsic reasoning within a closed-loop environment driven by RAGES (Retrieval-Augmented Generation-based Examination Simulator) . Our training methodology uses a tri-factor reward design to enforce clinical rigor: (1) a format reward to  \nCurated Pathological Corpus  \nA 79year old woman presented with a small gastric antralpolyp.  \nThe image shows a mucosa and submucosal multinodular proliferation of cytologically bland spindle cells with a myxoid background. Mitoses were not identified.  \nThe possible diagnoses includes gastric plexiformfibromyxoma, inflammatory fibroidpolyp and GIST.  \nAdditional test ofKi67 demonstrates a very ","cbCaidypcfP3J1mo","https://ap.wps.com/l/cbCaidypcfP3J1mo","pdf",5047260,33,"English","# Abstract\n# 1 Introduction\n# 2 Related Work\n## 2.1 Reasoning Ca","[{\"question\":\"How does the paper address the mismatch between LLM inference and real clinical workflows?\",\"answer\":\"It models medical diagnosis as an iterative evidence-seeking task rather than a one-shot inference problem, reflecting the active investigative nature of clinical practice.\"},{\"question\":\"What is RLVR used for in this framework?\",\"answer\":\"Reinforcement Learning with Verifiable Rewards (RLVR) is used to elicit intrinsic reasoning in a closed-loop environment, guided by reward signals that enforce diagnostic precision and consistency.\"},{\"question\":\"What role does RAGES play in the proposed system?\",\"answer\":\"RAGES is a retrieval-augmented examination simulator that acts as a high-fidelity clinical oracle, generating realistic follow-up evidence to support an autonomous reasoning loop.\"}]","Reinforcement Learning for Evidence-Seeking Diagnostic Reasoning with Large Language Models | PDF",1789977685,83]