[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82306-en":3,"doc-seo-82306-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},82306,1374391974564,"Clementine","https://ap-avatar.wpscdn.com/avatar/14000253aa45c000a9e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779874745381141002",8,"Research & Report","LongMedBench Benchmarking Medical Agents for Long Horizon Clinical Decision Making","LongMedBench introduces a real-world EHR-based benchmark for long-horizon clinical decision-making, addressing the gap where prior LLM medical agent evaluations focus on short-context QA and tool use. It converts MIMIC-IV admission records and clinical notes into time-series event streams and long-context memory datasets for multi-session agent-environment interaction. The benchmark includes 335 patients (19.72 inpatient visits per patient; 44.91 events per visit). It proposes an evaluation taxonomy covering fact-based QA, temporal reasoning, and long-horizon decision-making, showing improved retrieval via RAG and agent memory while highlighting decision-making reliance on immediate context.","arXiv :2607 .09322v 1 [ cs .AI] 10 Jul 2026  \nLongMedBench: Benchmarking Medical Agents for Long-Horizon Clinical Decision-Making  \nZihan Xu 1 * , Yanzhen Chen 1 * , Xiaocheng Zhang 1 * , Zhiting Fan 1 , Weiqi Zhai2 ,  \nHongxia Xu 1 ,3 , and Zuozhu Liu 1 ,3†  \n1 Zhejiang University, Hangzhou, China  \n{zihan1.22,yanzhen.22,xiaocheng.22,[zhiting.23}@intl.zju.edu.cn](zhiting.23}@intl.zju.edu.cn)  \n[einstein@zju.edu.cn](einstein@zju.edu.cn) , [zuozhuliu@intl.zju.edu.cn](zuozhuliu@intl.zju.edu.cn)  \n2 Alibaba Group, Hangzhou, China  \n[zhaiweiqi.zwq@alibaba-inc.com](zhaiweiqi.zwq@alibaba-inc.com)  \n3 Transvascular Implantation Devices Research Institute, Hangzhou, China  \nAbstract. In this work, we introduce LongMedBench, a real-world EHRbased benchmark for long-horizon clinical decision-making. Prior evaluations of LLM-based medical agents have largely emphasized shortcontext knowledge QA and tool use. However, real-world medical care is inherently longitudinal, and clinicians must aggregate evidence across repeated visits, tests, and evolving treatments. Therefore, long-horizon interaction is essential for realistic assessment. LongMedBench is constructed via a reproducible pipeline that integrates MIMIC-IV admission records and clinical notes into time-series event streams and long-context memory datasets, enabling long-horizon, multi-session interactions between agents and a clinical environment. It comprises 335 patients, with  \n19.72 inpatient visits per patient on average and 44.91 medical events per visit. Guided by the long-horizon decision process, we propose an evaluation taxonomy with three suites: fact-based QA, temporal reasoning, and long-horizon decision-making. This taxonomy measures how agents understand and leverage historical patient information over extended horizons. Our experiments show that while recent LLMs can make good use of explicit timestamps, they have challenges in implicit time inference; The RAG and agent memory system can improve the performance of information retrieval tasks, but the performance of decision-making tasks is highly dependent on the model’s immediate context.  \nKeywords: Computer-Aided Diagnosis · Medical Agents · EHR.  \n1 Introduction  \nIn clinical practice, the diagnostic process is inherently longitudinal and timedependent. Clinicians do not merely react to isolated symptoms; instead, they must synthesize evidence across multiple visits, diagnostic tests, and evolving  \n*  \nThese authors contributed equally to the work.  \n† Corresponding author: Zuozhu Liu.  \n2 Z. Xu et al.  \nTable 1: Comparison of Medical Agent Benchmarks.  \n\n| Benchmark | Long.a | Ctx.b | EHR.c | Dec.d | Temp.e | Focus |\n| --- | --- | --- | --- | --- | --- | --- |\n| MedAgentBench[6] | × | × | ✓ | ✓ | × | EHR Tool Integration |\n| AgentClinic[17] | × | × | × | ✓ | × | Simulated Interaction |\n| DiagBench[15] | × | × | ✓ | ✓ | × | Diagnostic Trajectory |\n| MedBench v4[3] | × | ✓ | ✓ | × | × | Medical Knowledge QA |\n| ReflecTool[10] | × | × | × | ✓ | × | Reflective Tool-Use |\n| EHRSQL[9] | × | × | ✓ | × | × | Relational Querying |\n| LongMedBench | ✓ | ✓ | ✓ | ✓ | ✓ | Long-horizon Reasoning |\n\na Longitudinal: Reasoning across multiple discrete clinical visits. b Context: Whether the benchmark environment contains multi-turn context. cEHR: The dataset built upon realworld EHR. d Decision-making: Evaluating proactive clinical planning beyond static retrieval or SQL-based querying. e Temporal Sensitivity: Evaluating understanding of clinical timeline and urgency.  \ntreatment responses over years [4] . This capacity of long-horizon clinical reasoning is fundamental to high-quality care. As Large Language Models (LLMs) transition to autonomous medical agents, their ability to navigate complex trajectories in real-world Electronic Health Records (EHR) [7] has become the critical benchmark for clinical readiness.  \nAlthough general-purpose benchmarks have emerged to evaluate long-horizon or multi-turn interactions [19, 18, 8","cbCairByc5RwqKiw","https://ap.wps.com/l/cbCairByc5RwqKiw","pdf",3752988,3,1,10,"English","en",105,"# Introduction\n## Problem and motivation\n## Benchmark overview\n## Contributions\n# Dataset construction\n## Memory architecture\n# Evaluation taxonomy\n## Fact-based QA\n## Temporal reasoning\n## Long-horizon decision-making\n# Experiments and results\n## Retrieval vs decision-making","[{\"question\":\"What limitation do LongMedBench and prior medical agent benchmarks address?\",\"answer\":\"LongMedBench targets the limitation that earlier evaluations mainly measure short-context question answering and tool use, while real clinical care requires longitudinal evidence aggregation across repeated visits and evolving treatments.\"},{\"question\":\"How is LongMedBench constructed from clinical data?\",\"answer\":\"It integrates MIMIC-IV admission records and clinical notes into time-series event streams and long-context memory datasets, enabling multi-session interaction between agents and a clinical environment.\"},{\"question\":\"What evaluation tasks are included in LongMedBench?\",\"answer\":\"LongMedBench uses a tri-level evaluation taxonomy with three suites: fact-based QA, temporal reasoning, and long-horizon decision-making, measuring how agents use historical patient information over extended horizons.\"}]",1784179502,25,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"longmedbench-benchmarking-medical-agents-for-long-horizon-clinical-decision-making","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":20},"https://docshare.wps.com/document/research-report/",{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/longmedbench-benchmarking-medical-agents-for-long-horizon-clinical-decision-making/82306/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-23","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What limitation do LongMedBench and prior medical agent benchmarks address?","Question",{"text":75,"@type":76},"LongMedBench targets the limitation that earlier evaluations mainly measure short-context question answering and tool use, while real clinical care requires longitudinal evidence aggregation across repeated visits and evolving treatments.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is LongMedBench constructed from clinical data?",{"text":80,"@type":76},"It integrates MIMIC-IV admission records and clinical notes into time-series event streams and long-context memory datasets, enabling multi-session interaction between agents and a clinical environment.",{"name":82,"@type":73,"acceptedAnswer":83},"What evaluation tasks are included in LongMedBench?",{"text":84,"@type":76},"LongMedBench uses a tri-level evaluation taxonomy with three suites: fact-based QA, temporal reasoning, and long-horizon decision-making, measuring how agents use historical patient information over extended horizons.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":22,"slug":133},"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]