An Adaptive Foundation Model with Evidence-based Clinical Reasoning for Gastroenterology

An Adaptive Foundation Model with Evidence-based Clinical Reasoning for Gastroenterology | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article An Adaptive Foundation Model with Evidence-based Clinical Reasoning for Gastroenterology Yixuan Yuan, Wenting Chen, Shengyuan Liu, Boyun Zheng, Jipeng Zhang, and 7 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8459685/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Gastrointestinal diseases affect 2.86 billion people globally, with capsule endoscopy (CE) providing crucial diagnostics but requiring manual review of over 60,000 frames per examination, a process associated with 17.4% disease miss rates. While artificial intelligence shows promise for CE analysis, existing endoscopic vision-language models (VLMs) lack multi-video understanding capability and cannot replicate the systematic multi-evidence reasoning that gastroenterologists integrate findings across anatomical regions to synthesize cohesive diagnoses. Here we introduce CE-R1, an adaptive foundation model with evidence-based clinical reasoning capabilities specifically designed for gastroenterology. CE-R1 incorporates a dynamic router that assesses query complexity and selectively routes cases to either a lightweight model for straightforward questions or a deep reasoning model that generates transparent, step-by-step diagnostic thought processes. To enable this capability, we construct CE-Bench, the first large-scale multimodal CE dataset comprising 502,066 visual question-answering pairs with chain-of-thought reasoning annotations, spanning 70 fine-grained clinical sub-tasks across five core diagnostic categories: anatomy identification, endoscopic findings recognition, disease diagnosis, treatment planning, and medical report generation. Comprehensive evaluation on both in-distribution and out-of-distribution datasets from four independent hospitals demonstrates that CE-R1 achieves 86.7% overall accuracy, substantially outperforming state-of-the-art VLMs (best baseline: 24.6%) and surpassing average physician performance (39.9%) by 21.1%. CE-R1 maintains superior generalization across external validation sets (65.1–81.9% accuracy). Critically, the multi-evidence clinical reasoning capability delivers substantial performance gains in complex diagnostic tasks: CE-R1 surpasses the model without reasoning by 8.5% in disease diagnosis, demonstrating the clinical value of transparent, step-by-step diagnostic processes. These results establish CE-R1 as a robust foundation model for comprehensive CE analysis with immediate applications in clinical decision support and medical education. Health sciences/Gastroenterology/Gastrointestinal diseases Health sciences/Diseases/Gastrointestinal diseases Full Text Additional Declarations There is NO Competing Interest. Supplementary Files supplement.pdf Supplementary information Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8459685","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":577805945,"identity":"45ffc7b5-a56b-4303-940f-def0bcfe406f","order_by":0,"name":"Yixuan Yuan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6UlEQVRIiWNgGAWjYPACGzDJzHAASB5gYGNsIKwljXQth0nQYnAjx0zi547zcvwMzM+kC84wyPHdSGB7OAOfljNnzCR7z9w2lmxgM5OecYPBWPJGArvhBnxajveYSfC23U7ccICHTZrnA0PiBqAtkg/waTnMYyb5t+1c/X6olnrCWoC2SPO2HUgwYABpucGQYADSgs9hkmeOFVvLtiUbzjjMZmzNc0bCcOaZh22S+LzPdyN54823bXby/O3ND2/zHLOR5zuefEyyB48WhQMcBhAWM5iUAGICESnfwP4Ar4JRMApGwSgYBQwA0rpN6xVcZhwAAAAASUVORK5CYII=","orcid":"","institution":"The Chinese University of Hong Kong","correspondingAuthor":true,"prefix":"","firstName":"Yixuan","middleName":"","lastName":"Yuan","suffix":""},{"id":577805946,"identity":"24bda56f-ea67-4450-986f-5efad1b34353","order_by":1,"name":"Wenting Chen","email":"","orcid":"https://orcid.org/0000-0002-7457-9540","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Wenting","middleName":"","lastName":"Chen","suffix":""},{"id":577805947,"identity":"4e895d69-1699-4f65-9765-0d37a95d2904","order_by":2,"name":"Shengyuan Liu","email":"","orcid":"","institution":"The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Shengyuan","middleName":"","lastName":"Liu","suffix":""},{"id":577805948,"identity":"05d8c940-d6cc-494b-b3e1-2fd9e6351517","order_by":3,"name":"Boyun Zheng","email":"","orcid":"","institution":"The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Boyun","middleName":"","lastName":"Zheng","suffix":""},{"id":577805949,"identity":"ea11bd5b-5a37-444f-84cd-d92610fd6f56","order_by":4,"name":"Jipeng Zhang","email":"","orcid":"","institution":"The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Jipeng","middleName":"","lastName":"Zhang","suffix":""},{"id":577805950,"identity":"08c3800e-7116-4e55-a2a3-de947b136e8f","order_by":5,"name":"Wenxuan Wang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Wenxuan","middleName":"","lastName":"Wang","suffix":""},{"id":577805951,"identity":"a2113e8e-330c-4652-9bea-12a1fc2be18e","order_by":6,"name":"Dejun Fan","email":"","orcid":"","institution":"The Sixth Affiliated Hospital, Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Dejun","middleName":"","lastName":"Fan","suffix":""},{"id":577805952,"identity":"8185e63d-7797-488b-baff-6751d030d986","order_by":7,"name":"Raymond Tang","email":"","orcid":"","institution":"The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Raymond","middleName":"","lastName":"Tang","suffix":""},{"id":577805953,"identity":"e5d9f983-a6b4-4ff2-a2bf-3c4d0d89e331","order_by":8,"name":"Thomas Yuen Tung Lam","email":"","orcid":"https://orcid.org/0000-0002-4306-4990","institution":"The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"Yuen Tung","lastName":"Lam","suffix":""},{"id":577805954,"identity":"8884b781-bddd-4fb7-845e-c2bb57e06b37","order_by":9,"name":"Shannon Melissa Chan","email":"","orcid":"https://orcid.org/0000-0001-6280-4176","institution":"Department of Surgery, The Chinese University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Shannon","middleName":"Melissa","lastName":"Chan","suffix":""},{"id":577805955,"identity":"a5442284-9dd8-40b1-8f1d-936a4eb0872b","order_by":10,"name":"Lei Xing","email":"","orcid":"https://orcid.org/0000-0003-2536-5359","institution":"Stanford Medicine","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Xing","suffix":""},{"id":577805956,"identity":"dc2243dc-aef2-4903-8931-f5002cc0a930","order_by":11,"name":"Jiancong Hu","email":"","orcid":"","institution":"The Sixth Affiliated Hospital, Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Jiancong","middleName":"","lastName":"Hu","suffix":""}],"badges":[],"createdAt":"2025-12-27 08:35:07","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8459685/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8459685/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100826492,"identity":"005adf06-caac-4b54-980d-b53bdaa53cc5","added_by":"auto","created_at":"2026-01-21 19:21:33","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10160985,"visible":true,"origin":"","legend":"","description":"","filename":"CER1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8459685/v1/13c646372e06f71692081332.pdf"},{"id":100858544,"identity":"89d41375-385b-4754-892a-e97d644a99f1","added_by":"auto","created_at":"2026-01-22 07:24:25","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13155,"visible":true,"origin":"","legend":"","description":"","filename":"NCOMMS25105003.json","url":"https://assets-eu.researchsquare.com/files/rs-8459685/v1/2cc48144a659efdb99bdec08.json"},{"id":100826491,"identity":"807d63da-05bc-4f62-b82b-dd289d9dc0e0","added_by":"auto","created_at":"2026-01-21 19:21:33","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":99227,"visible":true,"origin":"","legend":"","description":"","filename":"supplement.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8459685/v1/70913a82706cbcc0e07bad7f.pdf"},{"id":104781254,"identity":"c36a0daa-69b3-400e-babb-253361b6578a","added_by":"auto","created_at":"2026-03-17 07:55:14","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4743736,"visible":true,"origin":"","legend":"Article File","description":"","filename":"CER1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8459685/v1_covered_f2f30b40-14ae-480c-91b4-1c79db2eb2ae.pdf"},{"id":100826490,"identity":"f121a176-6661-4c29-9c03-8f822479d9ef","added_by":"auto","created_at":"2026-01-21 19:21:33","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":99227,"visible":true,"origin":"","legend":"Supplementary information","description":"","filename":"supplement.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8459685/v1/9c995e576420a8798714c4b2.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"An Adaptive Foundation Model with Evidence-based Clinical Reasoning for Gastroenterology","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8459685/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8459685/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Gastrointestinal diseases affect 2.86 billion people globally, with capsule endoscopy (CE) providing crucial diagnostics but requiring manual review of over 60,000 frames per examination, a process associated with 17.4% disease miss rates. While artificial intelligence shows promise for CE analysis, existing endoscopic vision-language models (VLMs) lack multi-video understanding capability and cannot replicate the systematic multi-evidence reasoning that gastroenterologists integrate findings across anatomical regions to synthesize cohesive diagnoses. Here we introduce CE-R1, an adaptive foundation model with evidence-based clinical reasoning capabilities specifically designed for gastroenterology. CE-R1 incorporates a dynamic router that assesses query complexity and selectively routes cases to either a lightweight model for straightforward questions or a deep reasoning model that generates transparent, step-by-step diagnostic thought processes. To enable this capability, we construct CE-Bench, the first large-scale multimodal CE dataset comprising 502,066 visual question-answering pairs with chain-of-thought reasoning annotations, spanning 70 fine-grained clinical sub-tasks across five core diagnostic categories: anatomy identification, endoscopic findings recognition, disease diagnosis, treatment planning, and medical report generation. Comprehensive evaluation on both in-distribution and out-of-distribution datasets from four independent hospitals demonstrates that CE-R1 achieves 86.7% overall accuracy, substantially outperforming state-of-the-art VLMs (best baseline: 24.6%) and surpassing average physician performance (39.9%) by 21.1%. CE-R1 maintains superior generalization across external validation sets (65.1–81.9% accuracy). Critically, the multi-evidence clinical reasoning capability delivers substantial performance gains in complex diagnostic tasks: CE-R1 surpasses the model without reasoning by 8.5% in disease diagnosis, demonstrating the clinical value of transparent, step-by-step diagnostic processes. These results establish CE-R1 as a robust foundation model for comprehensive CE analysis with immediate applications in clinical decision support and medical education.","manuscriptTitle":"An Adaptive Foundation Model with Evidence-based Clinical Reasoning for Gastroenterology","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-21 19:21:28","doi":"10.21203/rs.3.rs-8459685/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"31586f9b-e513-475a-a771-ea5c5e0f2a3f","owner":[],"postedDate":"January 21st, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":61473802,"name":"Health sciences/Gastroenterology/Gastrointestinal diseases"},{"id":61473803,"name":"Health sciences/Diseases/Gastrointestinal diseases"}],"tags":[],"updatedAt":"2026-03-12T16:02:08+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-21 19:21:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8459685","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8459685","identity":"rs-8459685","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}