PRIME-Care: A Unified Reinforcement Learning and Mathematical Optimization Framework for Personalized Treatment Planning Under Clinical Uncertainty in Telemedicine

PRIME-Care: A Unified Reinforcement Learning and Mathematical Optimization Framework for Personalized Treatment Planning Under Clinical Uncertainty in Telemedicine | 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 PRIME-Care: A Unified Reinforcement Learning and Mathematical Optimization Framework for Personalized Treatment Planning Under Clinical Uncertainty in Telemedicine Mazyar Taghavi, Mina Mohammadi, Ihsan Ullah, Javad Vahidi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8390633/v2 This work is licensed under a CC BY 4.0 License Status: Posted Version 2 posted You are reading this latest preprint version Show more versions Abstract This study presents PRIME-Care, a unified framework combining reinforcement learning (RL) and mathematical optimization for personalized treatment planning under clinical uncertainty. The framework aims to improve the safety, personalization, and predictive accuracy of treatment strategies in dynamic, uncertain healthcare environments. PRIME-Care incorporates a hierarchical bilevel structure where the optimization layer enforces safety constraints, and the RL layer adapts treatment policies based on patient-specific dynamics. Key innovations include the propagation of uncertainty through the treatment planning process and the use of probabilistic latent-state models for more accurate disease progression forecasting. Experimental results demonstrate that PRIMECare outperforms traditional RL models and optimization-based approaches in terms of safety, personalization, and predictive accuracy. Specifically, it reduces critical constraint violations by over 70%, exhibits superior trajectory alignment to clinician-curated plans, and provides more stable, temporally consistent disease forecasts. Additionally, robustness tests show that PRIME-Care maintains near-optimal performance under perturbations, while traditional RL models experience significant degradation. These results suggest that PRIME-Care offers a promising solution for integrating AI-driven decision support into clinical workflows, providing safer, more personalized, and interpretable treatment plans. All code implementations are publicly available to ensure full reproducibility and facilitate further research in this domain. Telemedicine Personalized Treatment Reinforcement Learning Mathematical Optimization Clinical Decision Support Full Text Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 2 posted You are reading this latest preprint version Show more versions 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-8390633","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":562774437,"identity":"1778b4fe-a9ff-4552-96b7-4ec844802b02","order_by":0,"name":"Mazyar Taghavi","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0002-2512-8014","institution":"Iran University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Mazyar","middleName":"","lastName":"Taghavi","suffix":""},{"id":562774438,"identity":"6779644e-0654-4c51-9b1e-e62958736d0d","order_by":1,"name":"Mina Mohammadi","email":"","orcid":"","institution":"Kashan University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Mina","middleName":"","lastName":"Mohammadi","suffix":""},{"id":562774439,"identity":"fa075985-b085-4637-8b2e-7597e53af3a0","order_by":2,"name":"Ihsan Ullah","email":"","orcid":"https://orcid.org/0000-0002-5204-2283","institution":"Yuan Ze University","correspondingAuthor":false,"prefix":"","firstName":"Ihsan","middleName":"","lastName":"Ullah","suffix":""},{"id":562774440,"identity":"c350f01a-aa0a-4697-9746-d2311c303ad9","order_by":3,"name":"Javad Vahidi","email":"","orcid":"https://orcid.org/0000-0002-6582-7493","institution":"Iran University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Javad","middleName":"","lastName":"Vahidi","suffix":""}],"badges":[],"createdAt":"2025-12-18 03:43:49","currentVersionCode":2,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8390633/v2","doiUrl":"https://doi.org/10.21203/rs.3.rs-8390633/v2","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99797309,"identity":"4c98feca-9ae2-4178-b3cd-531501b12069","added_by":"auto","created_at":"2026-01-08 13:45:36","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":521740,"visible":true,"origin":"","legend":"","description":"","filename":"PRIMECare.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8390633/v2_covered_c4758938-c460-405f-ba23-2b682290fba4.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"PRIME-Care: A Unified Reinforcement Learning\nand Mathematical Optimization Framework for\nPersonalized Treatment Planning Under Clinical\nUncertainty in Telemedicine","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Intelligent Knowledge City","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":"Telemedicine, Personalized Treatment, Reinforcement Learning, Mathematical Optimization, Clinical Decision Support","lastPublishedDoi":"10.21203/rs.3.rs-8390633/v2","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8390633/v2","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study presents PRIME-Care, a unified framework combining reinforcement learning (RL) and mathematical optimization for personalized treatment planning under clinical uncertainty. The framework aims to improve the safety, personalization, and predictive accuracy of treatment strategies in dynamic, uncertain healthcare environments. PRIME-Care incorporates a hierarchical bilevel structure where the optimization layer enforces safety constraints, and the RL layer adapts treatment policies based on patient-specific dynamics. Key innovations include the propagation of uncertainty through the treatment planning process and the use of probabilistic latent-state models for more accurate disease progression forecasting. Experimental results demonstrate that PRIMECare outperforms traditional RL models and optimization-based approaches in terms of safety, personalization, and predictive accuracy. Specifically, it reduces critical constraint violations by over 70%, exhibits superior trajectory alignment to clinician-curated plans, and provides more stable, temporally consistent disease forecasts. Additionally, robustness tests show that PRIME-Care maintains near-optimal performance under perturbations, while traditional RL models experience significant degradation. These results suggest that PRIME-Care offers a promising solution for integrating AI-driven decision support into clinical workflows, providing safer, more personalized, and interpretable treatment plans. All code implementations are publicly available to ensure full reproducibility and facilitate further research in this domain.\u003c/p\u003e","manuscriptTitle":"PRIME-Care: A Unified Reinforcement Learning\nand Mathematical Optimization Framework for\nPersonalized Treatment Planning Under Clinical\nUncertainty in Telemedicine","msid":"","msnumber":"","nonDraftVersions":[{"code":2,"date":"2026-01-07 19:08:39","doi":"10.21203/rs.3.rs-8390633/v2","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}},{"code":1,"date":"2025-12-19 07:06:00","doi":"10.21203/rs.3.rs-8390633/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":"bd67aec1-91c5-4813-9fc5-0d85a2bb98a5","owner":[],"postedDate":"January 7th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-19T07:06:00+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-07 19:08:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v2","identity":"rs-8390633","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8390633","identity":"rs-8390633","version":["v2"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}