MedThink: Enhancing Diagnostic Accuracy in Small Models via Teacher-Guided Reasoning Correction

MedThink: Enhancing Diagnostic Accuracy in Small Models via Teacher-Guided Reasoning Correction | 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 Research Article MedThink: Enhancing Diagnostic Accuracy in Small Models via Teacher-Guided Reasoning Correction Xinchun Su, Chunxu Luo, Lipeng Ma, Yixuan Li, Weidong Yang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9553270/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 Accurate clinical diagnosis requires extensive domain knowledge and complex clinical reasoning capabilities. Although large language models (LLMs) hold great potential for clinical reasoning, their high computational and memory requirements limit their deployment in resource-constrained environments. Knowledge distillation (KD) can compress LLM capabilities into smaller models, but traditional KD merely transfers superficial answer patterns and fails to preserve the structured reasoning required for reliable diagnosis. To address this, we propose a two-stage distillation framework, MedThink, designed to cultivate robust clinical reasoning in small language models (SLMs). In the first stage, a teacher LLM screens data and injects domain-knowledge explanations to fine-tune a student model, establishing a knowledge foundation. In the second stage, the teacher evaluates the student’s errors, generates reasoning chains linking knowledge to correct answers, and refines the student’s diagnostic reasoning through a second round of fine-tuning. We evaluate MedThink on general medical benchmarks and a gastroenterology dataset comprising 955 question-answer pairs. Experiments demonstrate that MedThink outperforms six distillation strategies in all benchmarks: achieving an improvement of up to 12.7% over the student baseline in general tasks, and reaching a total top accuracy of 56.4% in gastroenterology evaluation. This indicates that iterative distillation centered on reasoning can significantly enhance the diagnostic accuracy and generalization capabilities of SLMs whilst maintaining computational efficiency. Our code and data are publicly available at https://github.com/destinybird/PrecisionBoost. Artificial Intelligence and Machine Learning Knowledge Distillation Medicine Reasoning Large Language Models Full Text Additional Declarations The authors declare no competing interests. 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-9553270","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":631011413,"identity":"c4f633a7-a8a6-4ba4-a0ac-80912bd49632","order_by":0,"name":"Xinchun Su","email":"","orcid":"https://orcid.org/0009-0004-6934-823X","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Xinchun","middleName":"","lastName":"Su","suffix":""},{"id":631015992,"identity":"5b4daacc-f4a0-407b-a4d4-f98d25338a55","order_by":1,"name":"Chunxu Luo","email":"","orcid":"","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Chunxu","middleName":"","lastName":"Luo","suffix":""},{"id":631015995,"identity":"16b48a2e-274b-4174-ba52-7bdc63196f0b","order_by":2,"name":"Lipeng Ma","email":"","orcid":"","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Lipeng","middleName":"","lastName":"Ma","suffix":""},{"id":631015996,"identity":"bd01f0ba-db27-4e7a-81d4-1f69fe70b9c5","order_by":3,"name":"Yixuan Li","email":"","orcid":"","institution":"Fudan University","correspondingAuthor":false,"prefix":"","firstName":"Yixuan","middleName":"","lastName":"Li","suffix":""},{"id":631015997,"identity":"6da12d64-b976-4f62-b893-fe0b9d1ac097","order_by":4,"name":"Weidong Yang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2klEQVRIiWNgGAWjYFCCBBBhww9lMBOtJU2ygVQth0nQYnA8x/Bzwa/zEgbHk589YKiwTmxgP3sAv5Yzb4ylZ/bdljA488zcgOFMemIDT14CXi1mN3IMpHl7btcZ3Egwk2BsO5zYIMFjQEiL8W/ennMSBjfSv0kw/iNOi5k0z48DQC05QFsaiNBif+ZZmTVvQ7KE5Jk3ZRIJx9KN23hy8GuRbE/efJvnj50E3/H0bRIfaqxl+9nP4NfCwMBhwMDYBmUnADEbAfVAwP6AgeEPYWWjYBSMglEwggEAowJITn7AgFoAAAAASUVORK5CYII=","orcid":"","institution":"Fudan University","correspondingAuthor":true,"prefix":"","firstName":"Weidong","middleName":"","lastName":"Yang","suffix":""}],"badges":[],"createdAt":"2026-04-28 11:18:58","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9553270/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9553270/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108083684,"identity":"f98e193b-decc-4986-a156-864922fd415a","added_by":"auto","created_at":"2026-04-29 08:11:59","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":402807,"visible":true,"origin":"","legend":"","description":"","filename":"file.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9553270/v1_covered_5c2f379e-29f1-45d8-906b-590c568ff701.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eMedThink: Enhancing Diagnostic Accuracy in Small Models via Teacher-Guided Reasoning Correction\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Fudan University","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":"Knowledge Distillation, Medicine Reasoning, Large Language Models","lastPublishedDoi":"10.21203/rs.3.rs-9553270/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9553270/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurate clinical diagnosis requires extensive domain knowledge and complex clinical reasoning capabilities. Although large language models (LLMs) hold great potential for clinical reasoning, their high computational and memory requirements limit their deployment in resource-constrained environments. Knowledge distillation (KD) can compress LLM capabilities into smaller models, but traditional KD merely transfers superficial answer patterns and fails to preserve the structured reasoning required for reliable diagnosis.\u003c/p\u003e\n\u003cp\u003eTo address this, we propose a two-stage distillation framework, MedThink, designed to cultivate robust clinical reasoning in small language models (SLMs). In the first stage, a teacher LLM screens data and injects domain-knowledge explanations to fine-tune a student model, establishing a knowledge foundation. In the second stage, the teacher evaluates the student’s errors, generates reasoning chains linking knowledge to correct answers, and refines the student’s diagnostic reasoning through a second round of fine-tuning.\u003c/p\u003e\n\u003cp\u003eWe evaluate MedThink on general medical benchmarks and a gastroenterology dataset comprising 955 question-answer pairs. Experiments demonstrate that MedThink outperforms six distillation strategies in all benchmarks: achieving an improvement of up to 12.7% over the student baseline in general tasks, and reaching a total top accuracy of 56.4% in gastroenterology evaluation. This indicates that iterative distillation centered on reasoning can significantly enhance the diagnostic accuracy and generalization capabilities of SLMs whilst maintaining computational efficiency.\u003c/p\u003e\n\u003cp\u003eOur code and data are publicly available at https://github.com/destinybird/PrecisionBoost.\u003c/p\u003e","manuscriptTitle":"MedThink: Enhancing Diagnostic Accuracy in Small Models via Teacher-Guided Reasoning Correction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-29 08:11:19","doi":"10.21203/rs.3.rs-9553270/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":"76544d49-7c97-40f3-87b0-560108e6c05f","owner":[],"postedDate":"April 29th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67163187,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-04-29T08:11:19+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-29 08:11:19","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9553270","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9553270","identity":"rs-9553270","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}