GenAI Exceeds Clinical Experts in Predicting Acute Kidney Injury following Paediatric Cardiopulmonary Bypass

GenAI Exceeds Clinical Experts in Predicting Acute Kidney Injury following Paediatric Cardiopulmonary Bypass | 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 GenAI Exceeds Clinical Experts in Predicting Acute Kidney Injury following Paediatric Cardiopulmonary Bypass Alireza Mahani, Mansour Sharabiani, Alex Bottle, Yadav Srinivasan, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5370136/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 01 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted 6 You are reading this latest preprint version Abstract The emergence of large language models (LLMs) opens new horizons to leverage, often unused, information in clinical text. Our study aims to capitalise on this new potential. Specifically, we examine the utility of text embeddings generated by LLMs in predicting postoperative acute kidney injury (AKI) in paediatric cardiopulmonary bypass (CPB) patients using electronic health record (EHR) text, and propose methods for explaining their output. AKI could be a serious complication in paediatric CPB and its accurate prediction can significantly improve patient outcomes by enabling timely interventions. We evaluate various text embedding algorithms such as Doc2Vec, top-performing sentence transformers on Hugging Face, and commercial LLMs from Google and OpenAI. We benchmark the cross-validated performance of these 'AI models' against a 'baseline model' as well as an established clinically-defined 'expert model'. The baseline model includes structured features, i.e., patient gender, age, height, body mass index and length of operation. The majority of AI models surpass, not only the baseline model, but also the expert model. An ensemble of AI and clinical-expert models improves discriminative performance by 23% compared to the baseline model. Consistency of patient clusters formed from AI-generated embeddings with clinical-expert clusters - measured via the adjusted rand index and adjusted mutual information metrics - illustrates the medical validity of LLM embeddings. We create a reverse mapping from the numeric embedding space to the natural-language domain via the embedding-based clusters, generating medical labels for the clusters in the process. We also use text-generating LLMs to summarise the differences between AI and expert clusters. Such 'explainability' outputs can increase medical practitioners' trust in the AI applications, and help generate new hypotheses, e.g., by studying the association of cluster memberships and outcomes of interest. Health sciences/Diseases/Kidney diseases/Acute kidney injury Physical sciences/Mathematics and computing/Statistics generative artificial intelligence text embedding electronic health records cardiopulmonary bypass acute kidney injury spherical k-means Full Text Additional Declarations No competing interests reported. Supplementary Files clustermappingfullresults.csv supplementarytablelegends.txt AKILLMScientificReportsSuppMatrevisionminorv2.pdf Cite Share Download PDF Status: Published Journal Publication published 01 Jul, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Accepted 28 May, 2025 Reviews received at journal 11 May, 2025 Reviewers agreed at journal 03 May, 2025 Reviewers invited by journal 28 Apr, 2025 Submission checks completed at journal 28 Apr, 2025 First submitted to journal 09 Apr, 2025 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-5370136","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":449275722,"identity":"e5d8a295-d11a-4724-85c9-bdec6dcdf480","order_by":0,"name":"Alireza Mahani","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEklEQVRIiWNgGAWjYDACdsYGOJsZiHn4QayEAjxamNG1SIL4CQb4tKCzDQ6ASdw6+JuZ2z58+GPHIN/Anfi5oOaOjPH51YkfHhgwyPOLHcCqReIwY/PMGTzJQMN5N0vPOPaMx+zG280SQIcZzpydgN0aoBZmHglmoEt4N0jzsB0Gajm7AaQlweA2di3yIC1/DOqBDuPd/Jvn32Ee4xlnN//Ap8UApIUh4TADwwHebdK8bYd5DPh7t+G1xRCohbHnwHEeg8O826x5+w7zSNzg3WaRYCCB0y9yx9sfM/z4Uy0n3967+TbPt8P2/P1nN9/8UWEjzy+Nw/tQwIOIIAmwSgm8ytEA/wFSVI+CUTAKRsEIAAD8W1gsRPkb8gAAAABJRU5ErkJggg==","orcid":"","institution":"Statman Solution","correspondingAuthor":true,"prefix":"","firstName":"Alireza","middleName":"","lastName":"Mahani","suffix":""},{"id":449275723,"identity":"754c9a4c-f60a-4ced-af03-b2639b99420f","order_by":1,"name":"Mansour Sharabiani","email":"","orcid":"","institution":"Imperial College London, Department of Primary Care and Public Health","correspondingAuthor":false,"prefix":"","firstName":"Mansour","middleName":"","lastName":"Sharabiani","suffix":""},{"id":449275724,"identity":"21af4a70-5acb-416c-bab4-56be0dceeae5","order_by":2,"name":"Alex Bottle","email":"","orcid":"","institution":"Imperial College London, Department of Primary Care and Public Health","correspondingAuthor":false,"prefix":"","firstName":"Alex","middleName":"","lastName":"Bottle","suffix":""},{"id":449275725,"identity":"5bff5552-92a0-4c14-a12c-ded710bc2a1b","order_by":3,"name":"Yadav Srinivasan","email":"","orcid":"","institution":"Great Ormond Street Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yadav","middleName":"","lastName":"Srinivasan","suffix":""},{"id":449275726,"identity":"209593ec-68ec-4125-83d3-06e83d722f5c","order_by":4,"name":"Richard Issitt","email":"","orcid":"","institution":"Great Ormond Street Hospital","correspondingAuthor":false,"prefix":"","firstName":"Richard","middleName":"","lastName":"Issitt","suffix":""},{"id":449275727,"identity":"a573ab7b-b836-4599-b35e-6ce4db9e2c7f","order_by":5,"name":"Serban Stoica","email":"","orcid":"","institution":"Bristol Royal Hospital for Children","correspondingAuthor":false,"prefix":"","firstName":"Serban","middleName":"","lastName":"Stoica","suffix":""}],"badges":[],"createdAt":"2024-11-01 03:08:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5370136/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5370136/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-04651-8","type":"published","date":"2025-07-01T15:58:02+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":86179105,"identity":"5f5f8e10-fab2-4cc0-b9db-ae6642d5e155","added_by":"auto","created_at":"2025-07-07 16:15:45","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1420893,"visible":true,"origin":"","legend":"","description":"","filename":"AKILLMScientificReportsrevisionminorv2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5370136/v1_covered_ef27e5ed-4997-41f3-9457-66fefa623e38.pdf"},{"id":81607534,"identity":"543139ec-3d44-4aea-8135-995fa6c14218","added_by":"auto","created_at":"2025-04-29 06:11:28","extension":"csv","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3662,"visible":true,"origin":"","legend":"","description":"","filename":"clustermappingfullresults.csv","url":"https://assets-eu.researchsquare.com/files/rs-5370136/v1/1693fd9333888064181ab656.csv"},{"id":81606618,"identity":"184ab080-7c6b-491a-8e0c-af6bc29c8b3a","added_by":"auto","created_at":"2025-04-29 06:03:27","extension":"txt","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":536,"visible":true,"origin":"","legend":"","description":"","filename":"supplementarytablelegends.txt","url":"https://assets-eu.researchsquare.com/files/rs-5370136/v1/132536abc6bfe178a4e47f06.txt"},{"id":81607536,"identity":"747ceff5-9574-493e-98fc-3a01738f54aa","added_by":"auto","created_at":"2025-04-29 06:11:28","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":233746,"visible":true,"origin":"","legend":"","description":"","filename":"AKILLMScientificReportsSuppMatrevisionminorv2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5370136/v1/0d388f909ad6809b009b99e6.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"GenAI Exceeds Clinical Experts in Predicting Acute Kidney Injury following Paediatric Cardiopulmonary Bypass","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"generative artificial intelligence, text embedding, electronic health records, cardiopulmonary bypass, acute kidney injury, spherical k-means","lastPublishedDoi":"10.21203/rs.3.rs-5370136/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5370136/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The emergence of large language models (LLMs) opens new horizons to leverage, often unused, information in clinical text. Our study aims to capitalise on this new potential. Specifically, we examine the utility of text embeddings generated by LLMs in predicting postoperative acute kidney injury (AKI) in paediatric cardiopulmonary bypass (CPB) patients using electronic health record (EHR) text, and propose methods for explaining their output. AKI could be a serious complication in paediatric CPB and its accurate prediction can significantly improve patient outcomes by enabling timely interventions. We evaluate various text embedding algorithms such as Doc2Vec, top-performing sentence transformers on Hugging Face, and commercial LLMs from Google and OpenAI. We benchmark the cross-validated performance of these 'AI models' against a 'baseline model' as well as an established clinically-defined 'expert model'. The baseline model includes structured features, i.e., patient gender, age, height, body mass index and length of operation. The majority of AI models surpass, not only the baseline model, but also the expert model. An ensemble of AI and clinical-expert models improves discriminative performance by 23% compared to the baseline model. Consistency of patient clusters formed from AI-generated embeddings with clinical-expert clusters - measured via the adjusted rand index and adjusted mutual information metrics - illustrates the medical validity of LLM embeddings. We create a reverse mapping from the numeric embedding space to the natural-language domain via the embedding-based clusters, generating medical labels for the clusters in the process. We also use text-generating LLMs to summarise the differences between AI and expert clusters. Such 'explainability' outputs can increase medical practitioners' trust in the AI applications, and help generate new hypotheses, e.g., by studying the association of cluster memberships and outcomes of interest.","manuscriptTitle":"GenAI Exceeds Clinical Experts in Predicting Acute Kidney Injury following Paediatric Cardiopulmonary Bypass","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-29 06:03:23","doi":"10.21203/rs.3.rs-5370136/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Accepted","date":"2025-05-28T06:26:11+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-05-11T10:07:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"268586548309830530139966108222211065124","date":"2025-05-03T11:52:32+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-04-28T11:14:57+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-04-28T07:02:40+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-04-09T05:40:06+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"888a09d0-d127-4d64-bcd4-5665c10e31cb","owner":[],"postedDate":"April 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":47810018,"name":"Health sciences/Diseases/Kidney diseases/Acute kidney injury"},{"id":47810019,"name":"Physical sciences/Mathematics and computing/Statistics"}],"tags":[],"updatedAt":"2025-07-07T16:05:09+00:00","versionOfRecord":{"articleIdentity":"rs-5370136","link":"https://doi.org/10.1038/s41598-025-04651-8","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-07-01 15:58:02","publishedOnDateReadable":"July 1st, 2025"},"versionCreatedAt":"2025-04-29 06:03:23","video":"","vorDoi":"10.1038/s41598-025-04651-8","vorDoiUrl":"https://doi.org/10.1038/s41598-025-04651-8","workflowStages":[]},"version":"v1","identity":"rs-5370136","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5370136","identity":"rs-5370136","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}