We deploy ACIE (Agentic Clinical Information Extraction) at University Medicine Essen, an on-premise agentic RAG system that extracts structured data from complete patient contexts spanning hundreds of documents and thousands of FHIR resources. Clinicians configure what to extract through typed schemas, with no developer involvement.</p>\n<p>Key findings from our deployment on one of the largest clinical FHIR repositories in Europe (~2B resources, 1.7M patients):</p>\n<ul>\n<li>Across 7,326 clinician judgments in a retrospective lymphoma registry study, physicians accepted 96.5% of extractions, with per-type acceptance ranging from 80% to 99%</li>\n<li>We quantify the metadata gap in real clinical data: 56.5% of documents carry timestamps outside their encounter period, standard document-type codes cover only 2.4% of documents, and a third of documents are duplicates</li>\n<li>We trace how these data quality failures shaped architectural decisions, from bypassing encounter-based scoping to using agentic search over fixed retrieval pipelines</li>\n</ul>\n<p>The system runs entirely on-premise (Qwen 3.6 35B on 4xH100), with every extracted value grounded in cited source passages for clinician verification.</p>\n<p>We'd love to hear from others working on clinical IE in production, especially around data quality challenges in real hospital systems.</p>\n","updatedAt":"2026-06-19T15:06:39.026Z","author":{"_id":"64b7a2f52fbbea73ad91f78a","avatarUrl":"/avatars/bfc61e6664722bdb5264ec6bdb19bbe3.svg","fullname":"O. Çinar-Koraş","name":"osmalpkoras","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8782182335853577},"editors":["osmalpkoras"],"editorAvatarUrls":["/avatars/bfc61e6664722bdb5264ec6bdb19bbe3.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.19602","authors":[{"_id":"6a35597edb23715e9da12c11","name":"Osman Alperen Çinar-Koraş","hidden":false},{"_id":"6a35597edb23715e9da12c12","name":"Marie Bauer","hidden":false},{"_id":"6a35597edb23715e9da12c13","name":"Sameh Khattab","hidden":false},{"_id":"6a35597edb23715e9da12c14","name":"Merlin Engelke","hidden":false},{"_id":"6a35597edb23715e9da12c15","name":"Moon Kim","hidden":false},{"_id":"6a35597edb23715e9da12c16","name":"Stephan Settelmeier","hidden":false},{"_id":"6a35597edb23715e9da12c17","name":"Shigeyasu Sugawara","hidden":false},{"_id":"6a35597edb23715e9da12c18","name":"Fabian Freisleben","hidden":false},{"_id":"6a35597edb23715e9da12c19","name":"Felix Nensa","hidden":false},{"_id":"6a35597edb23715e9da12c1a","name":"Jens Kleesiek","hidden":false}],"publishedAt":"2026-06-17T00:00:00.000Z","submittedOnDailyAt":"2026-06-19T00:00:00.000Z","title":"Configurable Clinical Information Extraction with Agentic RAG: What Works, What Breaks, and Why","submittedOnDailyBy":{"_id":"64b7a2f52fbbea73ad91f78a","avatarUrl":"/avatars/bfc61e6664722bdb5264ec6bdb19bbe3.svg","isPro":false,"fullname":"O. Çinar-Koraş","user":"osmalpkoras","type":"user","name":"osmalpkoras"},"summary":"Patient contexts span hundreds of heterogeneous documents and thousands of structured data points, yet the document-level metadata that AI systems need for retrieval and triage is absent or incomplete. Standard retrieval-augmented generation fails on this data, mishandling temporal reasoning, cross-document dependencies, and missing metadata. We deploy ACIE (Agentic Clinical Information Extraction) at University Medicine Essen: an on-premise agentic RAG pipeline that reasons over complete patient contexts and grounds every answer in source passages for clinician verification. We quantify the metadata gap, trace the architectural decisions it shaped, and evaluate extraction alongside an independent retrospective lymphoma registry study, in which nuclear-medicine physicians verify every extracted value against its cited sources. Across 7,326 judgments, clinicians accepted 96.5\\% of extractions, with per-type acceptance ranging from 80\\% to 99\\%.","upvotes":0,"discussionId":"6a35597edb23715e9da12c1b","ai_summary":"ACIE, an agentic RAG system deployed in a clinical setting, demonstrates high accuracy in extracting medical information from complex patient contexts, achieving 96.5% acceptance rate by nuclear-medicine physicians across 7,326 judgments.","ai_keywords":["retrieval-augmented generation","agentic RAG pipeline","clinical information extraction","patient contexts","source passages","clinician verification","lymphoma registry study","nuclear-medicine physicians"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"64b7d3ce901f417d4867af42","name":"ikim-uk-essen","fullname":"IKIM","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/636a588537d9329b4a018a8f/7x0ebzBd0yREXd2uavdBr.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"organization":{"_id":"64b7d3ce901f417d4867af42","name":"ikim-uk-essen","fullname":"IKIM","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/636a588537d9329b4a018a8f/7x0ebzBd0yREXd2uavdBr.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.19602.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2606.19602 in a model README.md to link it from this page.

Cite arxiv.org/abs/2606.19602 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2606.19602 in a Space README.md to link it from this page.

Add this paper to a collection to link it from this page.