Ophtimus-V2-Tx: A Compact Domain-Specific LLM for Ophthalmic Diagnosis and Treatment Planning

Ophtimus-V2-Tx: A Compact Domain-Specific LLM for Ophthalmic Diagnosis and Treatment Planning | 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 Ophtimus-V2-Tx: A Compact Domain-Specific LLM for Ophthalmic Diagnosis and Treatment Planning Minwook Kwon, Kuk Jin Jang, Seung Ju Baek, Yong Seop Han, Hyonyoung Choi, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7191523/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 10 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted 15 You are reading this latest preprint version Abstract Recent advances in Large language models (LLMs) have shown promising results in clinical decision support, but they often fall short when applied to case-specific reasoning required in real-world medical practice. We present Ophtimus-V2-Tx, a lightweight, domain-specific Small LM (SLM) fine-tuned on over 10,000 ophthalmology case reports, built upon an 8B-parameter base model. The model is optimized to generate diagnosis and treatment suggestions aligned with standardized coding systems such as ICD-10-CM, ATC, and ICD-10-PCS. Using the CliBench framework, we evaluate its performance across four clinical tasks: primary diagnosis, secondary diagnosis, medication recommendation, and surgical procedure prediction. Ophtimus-V2- Tx achieved the highest full-code F1 score in medication (0.32) and surgical tasks (0.16), and outperformed GPT-4o in primary diagnosis (F1: 0.28 vs. 0.25). It also showed strong topic-specific performance, attaining 0.80 accuracy on multiple-choice questions related to Uveitis, a clinically complex condition. Our contributions include: (1) empirical validation of case-based fine-tuning for clinical task adaptation, (2) application of a hierarchical benchmarking framework for multi-dimensional evaluation, and (3) a reproducible pipeline for building efficient, deployable medical LLMs. These findings demonstrate the feasibility of compact, domain-adapted models in delivering competitive clinical performance. Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Health sciences/Health care Health sciences/Medical research Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 10 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 26 Aug, 2025 Reviews received at journal 22 Aug, 2025 Reviewers agreed at journal 21 Aug, 2025 Reviews received at journal 21 Aug, 2025 Reviews received at journal 13 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviews received at journal 11 Aug, 2025 Reviewers agreed at journal 05 Aug, 2025 Reviewers agreed at journal 31 Jul, 2025 Reviewers invited by journal 30 Jul, 2025 Editor invited by journal 29 Jul, 2025 Editor assigned by journal 25 Jul, 2025 Submission checks completed at journal 24 Jul, 2025 First submitted to journal 22 Jul, 2025 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-7191523","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":493843435,"identity":"e7f5883d-c685-4eaf-82fc-3636842759e3","order_by":0,"name":"Minwook Kwon","email":"","orcid":"","institution":"Gyeongsang National University","correspondingAuthor":false,"prefix":"","firstName":"Minwook","middleName":"","lastName":"Kwon","suffix":""},{"id":493843436,"identity":"392a0129-5d77-4ca3-b929-b6a110772d7d","order_by":1,"name":"Kuk Jin Jang","email":"","orcid":"","institution":"University of Pennsylvania","correspondingAuthor":false,"prefix":"","firstName":"Kuk","middleName":"Jin","lastName":"Jang","suffix":""},{"id":493843437,"identity":"c43e7cfa-6fce-4414-a756-37e9532c706a","order_by":2,"name":"Seung Ju Baek","email":"","orcid":"","institution":"Gyeongsang National University","correspondingAuthor":false,"prefix":"","firstName":"Seung","middleName":"Ju","lastName":"Baek","suffix":""},{"id":493843438,"identity":"bf021442-0874-48c8-ab7a-0e70737df622","order_by":3,"name":"Yong Seop Han","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yong","middleName":"Seop","lastName":"Han","suffix":""},{"id":493843439,"identity":"f2047d70-9037-45c4-96cf-233092d96cf3","order_by":4,"name":"Hyonyoung Choi","email":"","orcid":"","institution":"University of Pennsylvania","correspondingAuthor":false,"prefix":"","firstName":"Hyonyoung","middleName":"","lastName":"Choi","suffix":""},{"id":493843440,"identity":"b00b16ca-096f-458d-bb3b-55d543cdd420","order_by":5,"name":"Insup Lee","email":"","orcid":"","institution":"University of Pennsylvania","correspondingAuthor":false,"prefix":"","firstName":"Insup","middleName":"","lastName":"Lee","suffix":""},{"id":493843441,"identity":"dc26ae72-0892-458e-b082-3267eb75ceb9","order_by":6,"name":"Jin Hyun Kim","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxUlEQVRIiWNgGAWjYBACxmYgbjBgkIMJGBCtxZh4LWBtDQwMiQ1Ea2Fu5z3AOKNgW/r86B4Dhh81DMbmDQS0MDbzJTBuMLidu/HOGQPGnmMMZjIHCGrhMWB8ANIyI8eAgbeBwUaCoD+gWtINgVoY/xKtBeiwBHmJHANmoC1mRGk5OMPgtuEGibSCwzLHJIwJajHsP2P4sOfPbXn5GckbH76psTGcQVBLAwPDARDD4ACYQdAOBgZ5OKOBsOJRMApGwSgYoQAARUc6KcnFSwsAAAAASUVORK5CYII=","orcid":"","institution":"Gyeongsang National University","correspondingAuthor":true,"prefix":"","firstName":"Jin","middleName":"Hyun","lastName":"Kim","suffix":""}],"badges":[],"createdAt":"2025-07-23 03:08:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7191523/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7191523/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-27410-1","type":"published","date":"2025-12-10T15:59:21+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":98244811,"identity":"bbc48716-5902-46ae-a440-39479b0553d1","added_by":"auto","created_at":"2025-12-15 16:15:25","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2702267,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7191523/v1_covered_7dcb1355-bbe2-41a6-a518-48f8b18c490d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eOphtimus-V2-Tx: A Compact Domain-Specific LLM for Ophthalmic Diagnosis and Treatment Planning\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7191523/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7191523/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Recent advances in Large language models (LLMs) have shown promising results in clinical decision support, but they often fall short when applied to case-specific reasoning required in real-world medical practice. We present Ophtimus-V2-Tx, a lightweight, domain-specific Small LM (SLM) fine-tuned on over 10,000 ophthalmology case reports, built upon an 8B-parameter base model. The model is optimized to generate diagnosis and treatment suggestions aligned with standardized coding systems such as ICD-10-CM, ATC, and ICD-10-PCS. Using the CliBench framework, we evaluate its performance across four clinical tasks: primary diagnosis, secondary diagnosis, medication recommendation, and surgical procedure prediction. Ophtimus-V2- Tx achieved the highest full-code F1 score in medication (0.32) and surgical tasks (0.16), and outperformed GPT-4o in primary diagnosis (F1: 0.28 vs. 0.25). It also showed strong topic-specific performance, attaining 0.80 accuracy on multiple-choice questions related to Uveitis, a clinically complex condition. Our contributions include: (1) empirical validation of case-based fine-tuning for clinical task adaptation, (2) application of a hierarchical benchmarking framework for multi-dimensional evaluation, and (3) a reproducible pipeline for building efficient, deployable medical LLMs. These findings demonstrate the feasibility of compact, domain-adapted models in delivering competitive clinical performance.","manuscriptTitle":"Ophtimus-V2-Tx: A Compact Domain-Specific LLM for Ophthalmic Diagnosis and Treatment Planning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-04 18:19:11","doi":"10.21203/rs.3.rs-7191523/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-26T09:43:39+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-22T07:37:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"68767130182707233616264039059487847901","date":"2025-08-21T20:45:32+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-21T15:37:10+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-13T22:11:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"44189245648686267671091777543784115305","date":"2025-08-13T11:59:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"60837422618625309102854515519747550343","date":"2025-08-13T09:15:21+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-11T17:40:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"308447626764131469655847616907551702918","date":"2025-08-05T22:58:22+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"233543658670689800052420698698491360742","date":"2025-07-31T15:03:58+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-30T16:34:53+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-07-29T05:59:22+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-25T09:34:59+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-24T07:44:19+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-07-23T03:00:12+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f97b3207-bee5-47ca-a038-8cba8e3ce723","owner":[],"postedDate":"August 4th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":52631519,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":52631520,"name":"Health sciences/Diseases"},{"id":52631521,"name":"Health sciences/Health care"},{"id":52631522,"name":"Health sciences/Medical research"}],"tags":[],"updatedAt":"2025-12-15T16:09:38+00:00","versionOfRecord":{"articleIdentity":"rs-7191523","link":"https://doi.org/10.1038/s41598-025-27410-1","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-12-10 15:59:21","publishedOnDateReadable":"December 10th, 2025"},"versionCreatedAt":"2025-08-04 18:19:11","video":"","vorDoi":"10.1038/s41598-025-27410-1","vorDoiUrl":"https://doi.org/10.1038/s41598-025-27410-1","workflowStages":[]},"version":"v1","identity":"rs-7191523","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7191523","identity":"rs-7191523","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}