Physics-Informed AI Surrogates for Simulation-Driven Cardiac Digital Twins from Clinical MRI: Fibrosis-Driven Arrhythmia Prediction Across Five Cardiac Pathologies

Physics-Informed AI Surrogates for Simulation-Driven Cardiac Digital Twins from Clinical MRI: Fibrosis-Driven Arrhythmia Prediction Across Five Cardiac Pathologies | 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 Physics-Informed AI Surrogates for Simulation-Driven Cardiac Digital Twins from Clinical MRI: Fibrosis-Driven Arrhythmia Prediction Across Five Cardiac Pathologies Emmanuel Lwele, Francis Chikweto This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9477173/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract Background: Cardiac digital twins enable patient-specific modelling of electrophysiologybut remain computationally prohibitive and are typically restricted tospecialised centres with access to multimodal imaging, including lategadolinium enhancement (LGE) MRI for fibrosis characterisation. No scalable,open-source surrogate framework currently operates directly from the cinecardiac MRI data available in standard clinical practice. Methods: We present a simulation-driven framework that constructs patient-specificventricular geometries from the publicly available ACDC 2017 dataset, 100cine-MRI exams spanning five cardiac pathologies (normal, myocardialinfarction, dilated cardiomyopathy, hypertrophic cardiomyopathy, andabnormal right ventricle), and couples them with physiologically informedsynthetic fibrosis distributions to generate large-scale ground-truthelectrophysiological data. Three-dimensional myocardial meshes werereconstructed using the marching cubes algorithm and converted to \((k)\)-nearestneighbour graphs. Electrical activation was simulated via the eikonal equationsolved by Dijkstra's algorithm, producing local activation time (LAT) mapsunder apex and base pacing. A physics-informed graph neural network (PI-GNN)surrogate, enforcing the eikonal constraint as an auxiliary physics loss(\((\lambda = 0.1)\)), was trained on these simulation pairs. Results: The PI-GNN surrogate accurately reproduced eikonal activation maps across allfive pathological groups (mean \((R^2 = 0.924 \pm 0.018)\); normalisedMAE \((= 0.043 \pm 0.009)\)), reducing per-patient computation from\(({\sim}\,380)\),ms to \(({\sim}\,0.45)\),ms, a speedup of\(({\sim}\,850\times)\). Three-dimensional activation maps correctly localisedconduction delays to fibrotic border zones and qualitatively reproducedmacro-reentrant ventricular tachycardia (VT) circuit topology across allpathology groups. Ablation experiments confirmed that removing the eikonalphysics constraint reduced \((R^2)\) by 0.053, and removing fibrosis inputsreduced \((R^2)\) by 0.121. Conclusions: Full multimodal clinical datasets are not a prerequisite for clinically usefulcardiac digital twins. Combining widely accessible cine-MRI geometry withsimulation-augmented training and physics-informed learning enables scalable,near-real-time electrophysiological prediction across heterogeneous cardiacpathologies, providing a foundation for personalised arrhythmia riskstratification and ablation guidance. Health sciences/Cardiology Biological sciences/Computational biology and bioinformatics Physical sciences/Engineering Physical sciences/Mathematics and computing Health sciences/Medical research cardiac digital twin physics-informed neural network graph neural network eikonal equation arrhythmia prediction fibrosis modelling ventricular tachycardia ACDC dataset cardiac MRI surrogate model Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 19 May, 2026 Reviewers agreed at journal 06 May, 2026 Reviewers agreed at journal 06 May, 2026 Reviewers agreed at journal 29 Apr, 2026 Reviewers invited by journal 27 Apr, 2026 Editor assigned by journal 23 Apr, 2026 Submission checks completed at journal 23 Apr, 2026 First submitted to journal 20 Apr, 2026 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-9477173","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":635015457,"identity":"9a93c705-7986-4ed7-8a6a-2b31e19727a3","order_by":0,"name":"Emmanuel Lwele","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzUlEQVRIiWNgGAWjYFAC5gYGhgIGOQYJKF8Cn2IIYARqMWAwJl1LYgPRWvgZGNskPhjYpM+f3WPA8KOGIXFmAwEtkg2MbZIzDNJyN9w5Y8DYc4whcTYhWwwOMLbd5jE4nLtBIseAgbeBIXEeIS32IC1/DP6ny8/IMWD8S4wWA6BfbgPtSmC4kWPADLKFoMMkDjO2/+wxSDbccCOt4LDMMQljgt7nb28+bPCjwk5efkbyxodvamxkZxwgZA0zEvsAURE5CkbBKBgFo4AwAABI8jv3a7C0KQAAAABJRU5ErkJggg==","orcid":"","institution":"Sheffield Hallam University","correspondingAuthor":true,"prefix":"","firstName":"Emmanuel","middleName":"","lastName":"Lwele","suffix":""},{"id":635015459,"identity":"69f28159-ed14-4fa3-b92e-51b1eb760f90","order_by":1,"name":"Francis Chikweto","email":"","orcid":"","institution":"Tohoku University","correspondingAuthor":false,"prefix":"","firstName":"Francis","middleName":"","lastName":"Chikweto","suffix":""}],"badges":[],"createdAt":"2026-04-21 01:23:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9477173/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9477173/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108804909,"identity":"34fb7edc-8278-408d-8ffb-679067615994","added_by":"auto","created_at":"2026-05-08 15:24:13","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1601261,"visible":true,"origin":"","legend":"","description":"","filename":"PhysicsInformedAISurrogatesforSimulationDrivenCardiacDigitalTwinsfromClinicalMRIFibrosisDrivenArrhythmiaPrediction.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9477173/v1_covered_38236e1d-8b2c-4359-a9cb-f8a82534de8e.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Physics-Informed AI Surrogates for Simulation-Driven Cardiac Digital\n Twins from Clinical MRI: Fibrosis-Driven Arrhythmia Prediction\n Across Five Cardiac Pathologies","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"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":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"cardiac digital twin, physics-informed neural network, graph neural network, eikonal equation, arrhythmia prediction, fibrosis modelling, ventricular tachycardia, ACDC dataset, cardiac MRI, surrogate model","lastPublishedDoi":"10.21203/rs.3.rs-9477173/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9477173/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003eCardiac digital twins enable patient-specific modelling of electrophysiologybut remain computationally prohibitive and are typically restricted tospecialised centres with access to multimodal imaging, including lategadolinium enhancement (LGE) MRI for fibrosis characterisation. No scalable,open-source surrogate framework currently operates directly from the cinecardiac MRI data available in standard clinical practice.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003eWe present a simulation-driven framework that constructs patient-specificventricular geometries from the publicly available ACDC 2017 dataset, 100cine-MRI exams spanning five cardiac pathologies (normal, myocardialinfarction, dilated cardiomyopathy, hypertrophic cardiomyopathy, andabnormal right ventricle), and couples them with physiologically informedsynthetic fibrosis distributions to generate large-scale ground-truthelectrophysiological data. Three-dimensional myocardial meshes werereconstructed using the marching cubes algorithm and converted to \\((k)\\)-nearestneighbour graphs. Electrical activation was simulated via the eikonal equationsolved by Dijkstra's algorithm, producing local activation time (LAT) mapsunder apex and base pacing. A physics-informed graph neural network (PI-GNN)surrogate, enforcing the eikonal constraint as an auxiliary physics loss(\\((\\lambda = 0.1)\\)), was trained on these simulation pairs.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003eThe PI-GNN surrogate accurately reproduced eikonal activation maps across allfive pathological groups (mean \\((R^2 = 0.924 \\pm 0.018)\\); normalisedMAE \\((= 0.043 \\pm 0.009)\\)), reducing per-patient computation from\\(({\\sim}\\,380)\\),ms to \\(({\\sim}\\,0.45)\\),ms, a speedup of\\(({\\sim}\\,850\\times)\\). Three-dimensional activation maps correctly localisedconduction delays to fibrotic border zones and qualitatively reproducedmacro-reentrant ventricular tachycardia (VT) circuit topology across allpathology groups. Ablation experiments confirmed that removing the eikonalphysics constraint reduced \\((R^2)\\) by 0.053, and removing fibrosis inputsreduced \\((R^2)\\) by 0.121.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003eFull multimodal clinical datasets are not a prerequisite for clinically usefulcardiac digital twins. Combining widely accessible cine-MRI geometry withsimulation-augmented training and physics-informed learning enables scalable,near-real-time electrophysiological prediction across heterogeneous cardiacpathologies, providing a foundation for personalised arrhythmia riskstratification and ablation guidance.\u003c/p\u003e","manuscriptTitle":"Physics-Informed AI Surrogates for Simulation-Driven Cardiac Digital\n Twins from Clinical MRI: Fibrosis-Driven Arrhythmia Prediction\n Across Five Cardiac Pathologies","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-06 06:32:14","doi":"10.21203/rs.3.rs-9477173/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-19T13:48:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"4401299902548270406247688825417225594","date":"2026-05-06T14:43:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"325527176438212962000073403127791705436","date":"2026-05-06T14:30:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"189545744223789946736245388799123752743","date":"2026-04-29T14:07:45+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-27T13:38:07+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-24T00:17:12+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-23T04:47:18+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Digital Medicine","date":"2026-04-21T01:09:55+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a0de19b9-605f-4570-b850-aafac4493a7a","owner":[],"postedDate":"May 6th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-19T13:48:21+00:00","index":37,"fulltext":""},{"type":"reviewerAgreed","content":"4401299902548270406247688825417225594","date":"2026-05-06T14:43:25+00:00","index":35,"fulltext":""},{"type":"reviewerAgreed","content":"325527176438212962000073403127791705436","date":"2026-05-06T14:30:39+00:00","index":34,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67584924,"name":"Health sciences/Cardiology"},{"id":67584925,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":67584926,"name":"Physical sciences/Engineering"},{"id":67584927,"name":"Physical sciences/Mathematics and computing"},{"id":67584928,"name":"Health sciences/Medical research"}],"tags":[],"updatedAt":"2026-05-06T06:32:14+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-06 06:32:14","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9477173","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9477173","identity":"rs-9477173","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}