MEDITRON: Open Medical Foundation Models Adapted for Clinical Practice | 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 Biological Sciences - Article MEDITRON: Open Medical Foundation Models Adapted for Clinical Practice Zeming Chen, Angelika Romanou, Antoine Bonnet, Alejandro Hernández-Cano, and 36 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4139743/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Large language and multimodal models (LLMs and LMMs) will transform access to medical knowledge and clinical decision support. However, the current leading systems fall short of this promise, as they are either limited in scale, which restricts their capabilities, closed-source, which limits the extensions and scrutiny that can be applied to them, or not sufficiently adapted to clinical settings, which inhibits their practical use. In this work, we democratize large-scale medical AI systems by developing MEDITRON: a suite of open-source LLMs and LMMs with 7B and 70B parameters adapted to the medical domain. MEDITRON extends pretraining on a comprehensively curated medical corpus that includes biomedical literature and internationally recognized clinical practice guidelines. Evaluations using standard medical reasoning benchmarks show significant improvements over all current open-access models and several state-of-the-art commercial LLMs that are orders of magnitude larger, more expensive to host, and closed-source. Enhanced with visual processing capabilities, our MEDITRON-V model also outperforms all open-access models and much larger closed-source models on multimodal reasoning tasks for various biomedical imaging modalities. Beyond traditional benchmarks, we also create a novel and physician-driven adversarial question dataset grounded in real-world clinical settings, and a comprehensive 17-metric evaluation rubric to assess alignment and contextualization to real-world clinical practice. Applying this framework to MEDITRON-70B's responses, sixteen independent physicians found a high level of alignment across all metrics, including medical accuracy, safety, fairness, communication, and interpretation. The MEDITRON suite is a significant step forward in closing the technological gap between closed- and open-source medical foundation models. By releasing our methodologies, models, and real-world clinical practice benchmarks, we aim to drive the open-source development of more capable, representative, accessible, and transparent medical AI assistants. Health sciences/Medical research Health sciences/Health care large language model large multimodal model medical AI generative AI AI for health Full Text Additional Declarations There is NO Competing Interest. Supplementary Figures and Supplementary Tables are not available with this version. Cite Share Download PDF Status: Under Review 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-4139743","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Biological Sciences - Article","associatedPublications":[],"authors":[{"id":286702570,"identity":"105d6ff0-4d73-4ddc-aa14-259ae302b8b5","order_by":0,"name":"Zeming Chen","email":"","orcid":"https://orcid.org/0000-0002-2389-6968","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Zeming","middleName":"","lastName":"Chen","suffix":""},{"id":286702571,"identity":"3481d823-b62b-40b0-92ac-ea82eb9b3418","order_by":1,"name":"Angelika Romanou","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Angelika","middleName":"","lastName":"Romanou","suffix":""},{"id":286702572,"identity":"b486beea-3a65-4d20-b424-fb1e5433ace1","order_by":2,"name":"Antoine Bonnet","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Antoine","middleName":"","lastName":"Bonnet","suffix":""},{"id":286702573,"identity":"f5f184f1-4b91-4fec-8095-610bd2b0d648","order_by":3,"name":"Alejandro Hernández-Cano","email":"","orcid":"https://orcid.org/0009-0001-8224-6885","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Alejandro","middleName":"","lastName":"Hernández-Cano","suffix":""},{"id":286702575,"identity":"fe8513a6-4d27-46a3-abd2-2569574cd582","order_by":4,"name":"Badr Alkhamissi","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Badr","middleName":"","lastName":"Alkhamissi","suffix":""},{"id":286702577,"identity":"536de990-541b-4961-b90c-9a6141ac253e","order_by":5,"name":"Kyle Matoba","email":"","orcid":"","institution":"IDIAP","correspondingAuthor":false,"prefix":"","firstName":"Kyle","middleName":"","lastName":"Matoba","suffix":""},{"id":286702579,"identity":"1be914c8-0ebd-4858-818a-74573feae5e2","order_by":6,"name":"Francesco Salvi","email":"","orcid":"https://orcid.org/0009-0001-6884-6825","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Francesco","middleName":"","lastName":"Salvi","suffix":""},{"id":286702582,"identity":"7e980c0e-c660-429c-a3d4-4251e805e234","order_by":7,"name":"Matteo Pagliardini","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Matteo","middleName":"","lastName":"Pagliardini","suffix":""},{"id":286702584,"identity":"e74b4f81-41d4-4da7-bbb3-a1d2e7d0f9ca","order_by":8,"name":"Simin Fan","email":"","orcid":"","institution":"University of Michigan","correspondingAuthor":false,"prefix":"","firstName":"Simin","middleName":"","lastName":"Fan","suffix":""},{"id":286702585,"identity":"d695d9fe-d209-49e2-a28f-c4d1eb13e8ef","order_by":9,"name":"Andreas Köpf","email":"","orcid":"","institution":"Open-Assistant","correspondingAuthor":false,"prefix":"","firstName":"Andreas","middleName":"","lastName":"Köpf","suffix":""},{"id":286702586,"identity":"a373f312-5bae-4875-8154-3141527150d8","order_by":10,"name":"Amirkeivan Mohtashami","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Amirkeivan","middleName":"","lastName":"Mohtashami","suffix":""},{"id":286702587,"identity":"68b91608-b6f3-47d0-bb3d-deb119990820","order_by":11,"name":"Alexandre Sallinen","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Alexandre","middleName":"","lastName":"Sallinen","suffix":""},{"id":286702588,"identity":"e5daa49b-639a-4598-8995-d17c1ce280cd","order_by":12,"name":"Vinitra Swamy","email":"","orcid":"https://orcid.org/0000-0002-6840-5923","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Vinitra","middleName":"","lastName":"Swamy","suffix":""},{"id":286702589,"identity":"a75e79f6-fe03-41fb-af25-f65381df4b6e","order_by":13,"name":"Alireza Sakhaeirad","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Alireza","middleName":"","lastName":"Sakhaeirad","suffix":""},{"id":286702590,"identity":"f14c084f-bb2a-47f4-bb42-244a8262468e","order_by":14,"name":"Igor Krawczuk","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Igor","middleName":"","lastName":"Krawczuk","suffix":""},{"id":286702591,"identity":"91f7a7a3-e938-4afe-926d-1884a271479b","order_by":15,"name":"Deniz Bayazit","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Deniz","middleName":"","lastName":"Bayazit","suffix":""},{"id":286702592,"identity":"43e9e035-3b90-4870-a626-af58ff112d62","order_by":16,"name":"Axel Marmet","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Axel","middleName":"","lastName":"Marmet","suffix":""},{"id":286702595,"identity":"3af04031-a023-48d2-b6b9-f750d8fe2840","order_by":17,"name":"Li Mi","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Mi","suffix":""},{"id":286702597,"identity":"bd1fe8ea-b1db-49e5-a4c2-ca11864ce5b4","order_by":18,"name":"Noémie Boillat-Blanco","email":"","orcid":"","institution":"Lausanne University Hospital, Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Noémie","middleName":"","lastName":"Boillat-Blanco","suffix":""},{"id":286702599,"identity":"f60d12ce-b330-4a9e-ba3f-4da6327fcb9b","order_by":19,"name":"Kristina Keitel","email":"","orcid":"","institution":"Inselspital University Hospital, Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Kristina","middleName":"","lastName":"Keitel","suffix":""},{"id":286702601,"identity":"4d533212-e4bb-409c-8f7a-5e3bf644ee2f","order_by":20,"name":"Javier Elkin","email":"","orcid":"https://orcid.org/0000-0002-7645-0230","institution":"ICRC","correspondingAuthor":false,"prefix":"","firstName":"Javier","middleName":"","lastName":"Elkin","suffix":""},{"id":286702603,"identity":"138ede71-15a2-484e-998e-91674e93552b","order_by":21,"name":"Blaise Robert","email":"","orcid":"","institution":"ICRC","correspondingAuthor":false,"prefix":"","firstName":"Blaise","middleName":"","lastName":"Robert","suffix":""},{"id":286702604,"identity":"db120205-ed94-471d-a11f-7ae48685b2c6","order_by":22,"name":"Syrielle Montariol","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Syrielle","middleName":"","lastName":"Montariol","suffix":""},{"id":286702607,"identity":"aa26cc85-2085-48b8-8deb-da4095bd02c0","order_by":23,"name":"Silvia Bressan","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Silvia","middleName":"","lastName":"Bressan","suffix":""},{"id":286702608,"identity":"eeeba173-4e86-456c-bf10-e0a4e3842e15","order_by":24,"name":"David Chen","email":"","orcid":"https://orcid.org/0000-0001-8300-184X","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Chen","suffix":""},{"id":286702614,"identity":"b38c8930-96a1-4eea-af9d-761160b80068","order_by":25,"name":"Vincent Demers","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Vincent","middleName":"","lastName":"Demers","suffix":""},{"id":286702615,"identity":"26c295d0-4bb6-4aa8-9391-9832fc93fed4","order_by":26,"name":"Nina Emery","email":"","orcid":"https://orcid.org/0009-0001-5500-2797","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Nina","middleName":"","lastName":"Emery","suffix":""},{"id":286702616,"identity":"fced8fab-91a6-451e-aeb5-1f2edb28a0ea","order_by":27,"name":"Nicolas Glasson","email":"","orcid":"https://orcid.org/0009-0004-4706-2636","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Nicolas","middleName":"","lastName":"Glasson","suffix":""},{"id":286702617,"identity":"0914cfaf-96a9-45b6-ae7a-d88707f5e77b","order_by":28,"name":"Paulina Mensah","email":"","orcid":"https://orcid.org/0000-0002-0570-5662","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Paulina","middleName":"","lastName":"Mensah","suffix":""},{"id":286702618,"identity":"9564170f-fd96-4f7d-9e08-4c5d253ddcdf","order_by":29,"name":"Alix Miauton","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Alix","middleName":"","lastName":"Miauton","suffix":""},{"id":286702619,"identity":"a835ae71-a834-4b93-a157-8fbb9157b2b8","order_by":30,"name":"Ségolène Roemer","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Ségolène","middleName":"","lastName":"Roemer","suffix":""},{"id":286702620,"identity":"bd1ad078-46a5-4576-86df-6c8e51490fd9","order_by":31,"name":"Johan Siebert","email":"","orcid":"https://orcid.org/0000-0002-2871-3382","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Johan","middleName":"","lastName":"Siebert","suffix":""},{"id":286702621,"identity":"46b12371-6b1e-49c7-a326-8920005ac166","order_by":32,"name":"Carl Starvaggi","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Carl","middleName":"","lastName":"Starvaggi","suffix":""},{"id":286702622,"identity":"4f0a51e3-7aaa-4207-8c15-88ff51deacca","order_by":33,"name":"Véronique Suttels","email":"","orcid":"https://orcid.org/0000-0002-5261-0439","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Véronique","middleName":"","lastName":"Suttels","suffix":""},{"id":286702623,"identity":"ac0516ab-5a23-47f2-a8ba-614bcbcf85d7","order_by":34,"name":"Rainer Tan","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Rainer","middleName":"","lastName":"Tan","suffix":""},{"id":286702624,"identity":"e23b2192-617a-4409-9be7-ff0a1c755e0e","order_by":35,"name":"R. Taylor","email":"","orcid":"","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"R.","middleName":"","lastName":"Taylor","suffix":""},{"id":286702625,"identity":"f328e0c9-da30-4a33-8716-0f091d0506e3","order_by":36,"name":"Jacques du Toit","email":"","orcid":"https://orcid.org/0000-0002-0003-0168","institution":"Clinical Evaluation Group","correspondingAuthor":false,"prefix":"","firstName":"Jacques","middleName":"","lastName":"du Toit","suffix":""},{"id":286702627,"identity":"63a86170-5fa0-4d6b-9a09-bdec8dd86a70","order_by":37,"name":"Mary-Anne Hartley","email":"","orcid":"","institution":"EPFL, Yale University","correspondingAuthor":false,"prefix":"","firstName":"Mary-Anne","middleName":"","lastName":"Hartley","suffix":""},{"id":286702628,"identity":"d125374c-5322-4cb9-ad97-f7f11471212e","order_by":38,"name":"Martin Jaggi","email":"","orcid":"","institution":"EPFL","correspondingAuthor":false,"prefix":"","firstName":"Martin","middleName":"","lastName":"Jaggi","suffix":""},{"id":286702569,"identity":"e7f79667-e601-4546-9d54-6fdc6fc26a4d","order_by":39,"name":"Antoine Bosselut","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+0lEQVRIiWNgGAWjYLCCBCA2ALMKGBj4gdSHBxBBHIAZWYsBg4RkAwPjjARCWhiQtRgcIKCFf3b/wQcPd9gwmLO3P/7wweBwnfHx3oMNCQx2ebi0SNw5zGyQeCaNwbLnjJnkDIPDEmZnziUCtSQX43TXjWQ2icS2wwwGN3LYmHlAWm7kmD9IYDiQ2IBDhzxEy3+glvTHn/8AtRjPf2PYgE+LAUTLASAjwUCaAajFQIIHvxbDG8nGBoltyTxgv/QYpEvOOJMH9ItBMk4tcjcSHz782WYnBw6xHxXW/PztZw82fKiww6kFBnjQ2AYE1OPRPgpGwSgYBaOAgQEAl3Fa+CZW014AAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-8968-9649","institution":"EPFL","correspondingAuthor":true,"prefix":"","firstName":"Antoine","middleName":"","lastName":"Bosselut","suffix":""}],"badges":[],"createdAt":"2024-03-20 23:05:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4139743/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4139743/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":54317140,"identity":"fcc2b761-849b-4a38-98f4-11285a08227f","added_by":"auto","created_at":"2024-04-08 18:03:07","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2202159,"visible":true,"origin":"","legend":"Article File","description":"","filename":"Nature2024Meditron.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4139743/v1_covered_f00e0fef-d14b-4daa-8ce4-59445c289734.pdf"}],"financialInterests":"\u003cp\u003eThere is \u003cstrong\u003eNO\u003c/strong\u003e Competing Interest.\u003c/p\u003e\n\u003cp\u003eSupplementary Figures and Supplementary Tables are not available with this version.\u003c/p\u003e","formattedTitle":"MEDITRON: Open Medical Foundation Models Adapted for Clinical Practice","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"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":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"large language model, large multimodal model, medical AI, generative AI, AI for health","lastPublishedDoi":"10.21203/rs.3.rs-4139743/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4139743/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Large language and multimodal models (LLMs and LMMs) will transform access to medical knowledge and clinical decision support. However, the current leading systems fall short of this promise, as they are either limited in scale, which restricts their capabilities, closed-source, which limits the extensions and scrutiny that can be applied to them, or not sufficiently adapted to clinical settings, which inhibits their practical use. In this work, we democratize large-scale medical AI systems by developing MEDITRON: a suite of open-source LLMs and LMMs with 7B and 70B parameters adapted to the medical domain. MEDITRON extends pretraining on a comprehensively curated medical corpus that includes biomedical literature and internationally recognized clinical practice guidelines. Evaluations using standard medical reasoning benchmarks show significant improvements over all current open-access models and several state-of-the-art commercial LLMs that are orders of magnitude larger, more expensive to host, and closed-source. Enhanced with visual processing capabilities, our MEDITRON-V model also outperforms all open-access models and much larger closed-source models on multimodal reasoning tasks for various biomedical imaging modalities. Beyond traditional benchmarks, we also create a novel and physician-driven adversarial question dataset grounded in real-world clinical settings, and a comprehensive 17-metric evaluation rubric to assess alignment and contextualization to real-world clinical practice. Applying this framework to MEDITRON-70B's responses, sixteen independent physicians found a high level of alignment across all metrics, including medical accuracy, safety, fairness, communication, and interpretation. The MEDITRON suite is a significant step forward in closing the technological gap between closed- and open-source medical foundation models. By releasing our methodologies, models, and real-world clinical practice benchmarks, we aim to drive the open-source development of more capable, representative, accessible, and transparent medical AI assistants.","manuscriptTitle":"MEDITRON: Open Medical Foundation Models Adapted for Clinical Practice","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-03 18:46:54","doi":"10.21203/rs.3.rs-4139743/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"
[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"17535075-78eb-4676-b1ed-016e3fd9ea25","owner":[],"postedDate":"April 3rd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":30178071,"name":"Health sciences/Medical research"},{"id":30178072,"name":"Health sciences/Health care"}],"tags":[],"updatedAt":"2024-11-15T10:56:12+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-03 18:46:54","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4139743","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4139743","identity":"rs-4139743","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.