[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-395188-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-395188-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","physics-informed-generalizable-framework-bridges-the-sim-to-real-gap-in-ai-based-cerebrovascular-hemodynamics-modeling","Physics-Informed Generalizable Framework Bridges the Sim-to-Real Gap in AI-based Cerebrovascular Hemodynamics Modeling","","Accurate spatio-temporal cerebrovascular hemodynamics characterization is essential for neurovascular disease management, yet clinical translation is limited by a sim-to-real gap. Existing AI models trained on idealized in silico data struggle to generalize across the large topological diversity of patient-specific anatomy measured in vivo. 4DHemoX bridges this gap using 4DHemoDB and a physics-informed 4DHemoFormer that embeds Navier-Stokes constraints in attention and training, achieving robust cross-geometry generalization and efficient data-calibration with zero-shot and low in vivo data requirements.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/healthcare/","Healthcare",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/physics-informed-generalizable-framework-bridges-the-sim-to-real-gap-in-ai-based-cerebrovascular-hemodynamics-modeling/395188/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/physics-informed-generalizable-framework-bridges-the-sim-to-real-gap-in-ai-based-cerebrovascular-hemodynamics-modeling/395188.png","ImageObject",300,407,{"name":42,"@type":43},"Damian","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-30","2026-09-26",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",9,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does the document address in AI-based cerebrovascular hemodynamics modeling?","Question",{"text":63,"@type":64},"It addresses the sim-to-real gap that prevents models trained on idealized in silico data from adapting to the topological variance of patient-specific anatomy in vivo.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"What is 4DHemoX and what are its key components?",{"text":68,"@type":64},"4DHemoX is a physics-informed framework that includes 4DHemoDB, a hybrid dataset with in silico and in vivo 4D-flow MRI frames, and 4DHemoFormer, a physics-informed Transformer embedding Navier-Stokes constraints into attention and loss.",{"name":70,"@type":61,"acceptedAnswer":71},"How does the proposed approach evaluate performance and efficiency?",{"text":72,"@type":64},"It benchmarks against state-of-the-art models, showing robust cross-geometry generalization on unseen patient-specific structures, and demonstrates data efficiency by requiring only 10% of in vivo data for rapid calibration to clinical MRI measurements.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},395188,1790762856,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,110,115,119,122,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":108,"slug":109},7,40,"healthcare",{"id":111,"doc_module":4,"doc_module_name":25,"category_name":112,"show_sort_weight":113,"slug":114},8,"Research & Report",30,"research-report",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":116,"show_sort_weight":117,"slug":118},"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":107,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":139,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":144,"read_time":145,"preview_page_count":4},137451208677,"https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d","Physics-Informed Generalizable Framework Bridges the Sim-to-Real Gap in AI-based Cerebrovascular Hemodynamics Modeling  \nFen Miao  \nShenzhen Institute for Advanced Study, University of Electronic Science and Technology of China  \nChen Chen  \nThe Hong Kong Polytechnic University [https://orcid.org/0009-0007-6003-0948](https://orcid.org/0009-0007-6003-0948)  \nYuan Lin  \nShenzhen Institute for Advanced Study, University of Electronic Science and Technology of China  \nYun Zhang  \nSchool of Electronics and Communication Engineering, Sun Yat-sen University  \nYingjie He  \nShenzhen Institute for Advanced Study, University of Electronic Science and Technology of China  \nJiaqi Huang  \nShenzhen Institute for Advanced Study, University of Electronic Science and Technology of China  \nHonglei Zhao  \nFuwai Hospital, Chinese Academy of Medical Sciences·Shenzhen  \nJun Yang  \nUniversity of Electronic Science and Technology of China  \nHuiying Liang  \nIntelligent Medicine Institute, Fudan University  \nYe Li  \nShenzhen Institutes of Advanced Technology, Chinese Academy of Sciences [https://orcid.org/0000-](https://orcid.org/0000-)[ ](https://orcid.org/0000-)0002-5351-8546  \nJiannong Cao  \nHong Kong Polytechnic University  \nArticle  \nKeywords:  \nPosted Date: June 11th, 2026  \nDOI: [https://doi.org/10.21203/rs.3.rs-9923857/v1](https://doi.org/10.21203/rs.3.rs-9923857/v1)  \nLicense: 􀁱 􀅏 This work is licensed under a Creative Commons Attribution 4.0 International License.  \nRead Full License  \nAdditional Declarations: There is NO Competing Interest.  \n1 Physics-Informed Generalizable Framework Bridges  \n2 the Sim-to-Real Gap in AI-based Cerebrovascular  \n3 Hemodynamics Modeling  \n4 Fen Miao 1†, Chen Chen2,3†, Yuan Lin 1†, Yun Zhang4, Jiannong Cao2, Yingjie He 1, Jiaqi  \n5 Huang 1, Honglei Zhao5, Jun Yang 1*, Huiying Liang6*, Ye Li3,7*  \n6 1Shenzhen Institute for Advanced Study, University of Electronic Science and Technology of  \n7 China, Shenzhen, China  \n8 2Hong Kong Polytechnic University, Hong Kong, China  \n9 3Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, 10 China  \n11 4School of Electronics and Communication Engineering, Sun Yat-sen University, Shenzhen, 12 China  \n13 5Fuwai Hospital, Chinese Academy of Medical Sciences·Shenzhen, Shenzhen, China  \n14 6Intelligent Medicine Institute, Fudan University, Shanghai, China  \n15 7Shenzhen Institute of Advanced Technology, Shenzhen,China.  \n16 *Corresponding author. Email: [junyang@uestc.edu.cn](junyang@uestc.edu.cn) ; [lianghuiying@fudan.edu.cn](lianghuiying@fudan.edu.cn) ;  \n17 [ye.li@siat.ac.cn](ye.li@siat.ac.cn) ;  \n18 † These authors contributed equally to this work.  \n19  \n20 Abstract:  \n21 Accurate characterization of spatio-temporal cerebrovascular hemodynamics  \n22 is essential for neurovascular disease management. However, clinical  \n23 translation is severely constrained by the \"sim-to-real\" gap, largely because  \n24 existing models trained on idealized in silico data struggle to adapt to the 25 immense topological variance of patient-specific anatomies inherent in in vivo 26 measurements. Here, we present 4DHemoX, a physics-informed framework to 27 bridge this gap. Central to this framework is 4DHemoDB, a hybrid dataset 28 comprising 168,000 high-fidelity in silico frames across 280 diverse, 29 patient-specific vascular geometries and 800 in vivo 4D-flow MRI frames. To 30 process these highly irregular spatial graphs, we introduce 4DHemoFormer, a 31 physics-informed Transformer that embeds Navier-Stokes constraints directly 32 into its attention mechanism and loss landscape. Benchmarking against 33 state-of-the-art models (PINNs, FNO, DeepONet) demonstrates that 34 4DHemoFormer establishes robust cross-geometry generalization on unseen 35 patient-specific structures (nRMSE: 2.39 × 10⁻²). Building upon this foundation, 36 4DHemoFormer successfully bridges the sim-to-real gap. Beyond establishing 37 a competitive zero-shot baseline (nRMSE: 6.28 × 10⁻²), 4DHem","cbCaidTMeWLsfDvt","https://ap.wps.com/l/cbCaidTMeWLsfDvt","pdf",2297933,36,"English","# Abstract\n# Introduction","[{\"question\":\"What problem does the document address in AI-based cerebrovascular hemodynamics modeling?\",\"answer\":\"It addresses the sim-to-real gap that prevents models trained on idealized in silico data from adapting to the topological variance of patient-specific anatomy in vivo.\"},{\"question\":\"What is 4DHemoX and what are its key components?\",\"answer\":\"4DHemoX is a physics-informed framework that includes 4DHemoDB, a hybrid dataset with in silico and in vivo 4D-flow MRI frames, and 4DHemoFormer, a physics-informed Transformer embedding Navier-Stokes constraints into attention and loss.\"},{\"question\":\"How does the proposed approach evaluate performance and efficiency?\",\"answer\":\"It benchmarks against state-of-the-art models, showing robust cross-geometry generalization on unseen patient-specific structures, and demonstrates data efficiency by requiring only 10% of in vivo data for rapid calibration to clinical MRI measurements.\"}]","Physics-Informed Generalizable Framework Bridges the Sim-to-Real Gap in AI-based Cerebrovascular Hemodynamics Modeling | PDF",1790416752,91]