GPT-4o and the Quest for Machine Learning Interpretability in ICU Mortality Prediction

GPT-4o and the Quest for Machine Learning Interpretability in ICU Mortality Prediction | 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 GPT-4o and the Quest for Machine Learning Interpretability in ICU Mortality Prediction Moein Einollahzadeh Samadi, Kateryna Nikulina, Sebastian Johannes Fritsch, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4816139/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 13 Oct, 2025 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted 12 You are reading this latest preprint version Abstract Background Clinical utilization of machine learning is hampered by the lack of interpretability inherent in most non-linear black box modeling approaches, reducing trust among clinicians and regulators. Advanced large language models, such as GPT-4o, offer a potential framework for integrating medical knowledge into these models, potentially enhancing their interpretability. Methods Our study utilizes GPT-4o to generate detailed medical feature descriptions, which are aggregated into a comprehensive corpus and processed using TF-IDF vectorization. We then apply fuzzy C-means clustering to these vectorized features to identify significant mortality cause-specific feature clusters. A physician reviews the resulting clusters, validating their relevance to specific mortality causes in mechanically ventilated ICU patients. Subsequently, the resulting clusters inform the creation of weak mortality classifiers, which are combined into a strong classifier using boosting techniques, ultimately producing a GPT-enhanced boosting model for ICU mortality prediction. Results This study enrolled 16,018 mechanically ventilated ICU patients, divided into derivation (12,758) and validation (3,260) cohorts, to develop and evaluate a GPT-enhanced boosting model for predicting in-ICU mortality. Leveraging GPT-4o, we implemented an automated process for clustering mortality cause-specific features, resulting in six feature clusters: Liver Failure, Infection, Renal Failure, Hypoxia, Cardiac Failure, and Mechanical Ventilation. This approach significantly improved upon previous manual methods, automating the reconstruction of structured boosting models. While the GPT-enhanced model showed similar predictive accuracy to an XGBoost model, it demonstrated superior interpretability and clinical relevance by incorporating a wider array of features and providing a hierarchical structure of feature importance aligned with medical knowledge. Conclusion We introduce a novel approach to predicting in-ICU mortality for mechanically ventilated patients using a GPT-enhanced boosting model. Our methodology demonstrates the potential of integrating large language models with traditional machine learning techniques to create interpretable and clinically relevant predictive models. Interpretable machine learning Large language models Feature clustering ICU mortality prediction Full Text Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformation20240725MS.pdf Cite Share Download PDF Status: Published Journal Publication published 13 Oct, 2025 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted Editorial decision: Revision requested 28 Feb, 2025 Reviews received at journal 24 Feb, 2025 Reviewers agreed at journal 04 Feb, 2025 Reviews received at journal 21 Jan, 2025 Reviewers agreed at journal 10 Jan, 2025 Editor invited by journal 02 Dec, 2024 Reviewers agreed at journal 30 Sep, 2024 Reviewers agreed at journal 17 Aug, 2024 Reviewers invited by journal 15 Aug, 2024 Editor assigned by journal 01 Aug, 2024 Submission checks completed at journal 01 Aug, 2024 First submitted to journal 28 Jul, 2024 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-4816139","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":336674592,"identity":"3a90b16b-32f5-4e1c-9a9c-bd060f5fa8b5","order_by":0,"name":"Moein Einollahzadeh Samadi","email":"","orcid":"","institution":"Universitätsklinikum Aachen","correspondingAuthor":false,"prefix":"","firstName":"Moein","middleName":"Einollahzadeh","lastName":"Samadi","suffix":""},{"id":336674594,"identity":"0abd37b1-73fd-40fb-b17e-a6d1d8388473","order_by":1,"name":"Kateryna Nikulina","email":"","orcid":"","institution":"RWTH Aachen University","correspondingAuthor":false,"prefix":"","firstName":"Kateryna","middleName":"","lastName":"Nikulina","suffix":""},{"id":336674595,"identity":"89094010-9235-40a1-89f5-56ad93a9c38f","order_by":2,"name":"Sebastian Johannes Fritsch","email":"","orcid":"","institution":"Universitätsklinikum Aachen","correspondingAuthor":false,"prefix":"","firstName":"Sebastian","middleName":"Johannes","lastName":"Fritsch","suffix":""},{"id":336674597,"identity":"ad8da788-b09c-4e7a-84f2-d038c60667d0","order_by":3,"name":"Andreas Schuppert","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+UlEQVRIiWNgGAWjYHACNijNfICxAUgZgNg8RGlhY0sgWQuPAXFa+PsPP3vMu+OOnPn8nm+SM2rs5M3Zew8wvKnArUXiRpq5Me+ZZ8Yyx3i3SW44lmy4s+dcAuOcM3isucHDJs3bdjhxBhtQywM25gSDGzkGzLxtuHXInz8D1lI/g43nmeSDf/UJBvffALX8w63F4EAOWEuCBBsPm+RGIMPgBg9QSwNuLYY30swk5555ZjiDLc3YcmbfccMNZ3IMDs45hluL3PnDzyTe7rgjL8F8+OHNnm/V8gbHzxg+eFODx/sgwNhwAFXgADZV+LWMglEwCkbBKEAGAGgGUwWOsCiaAAAAAElFTkSuQmCC","orcid":"","institution":"Universitätsklinikum Aachen","correspondingAuthor":true,"prefix":"","firstName":"Andreas","middleName":"","lastName":"Schuppert","suffix":""}],"badges":[],"createdAt":"2024-07-28 09:56:26","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4816139/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4816139/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12911-025-03224-z","type":"published","date":"2025-10-13T15:58:08+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":93956779,"identity":"0afd3def-ce2f-430d-9d76-ee4ce1c7f22a","added_by":"auto","created_at":"2025-10-20 16:12:21","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4935021,"visible":true,"origin":"","legend":"","description":"","filename":"ManuscriptGPT420240731MS.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4816139/v1_covered_4e67c84d-6757-4de2-8440-477e350e6bba.pdf"},{"id":61890517,"identity":"103c1322-61e1-490a-b3f6-5f02d45229f7","added_by":"auto","created_at":"2024-08-06 18:15:52","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":122738,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation20240725MS.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4816139/v1/8ca9868d79b75a051e615c23.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"GPT-4o and the Quest for Machine Learning Interpretability in ICU Mortality Prediction","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":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Interpretable machine learning, Large language models, Feature clustering, ICU mortality prediction","lastPublishedDoi":"10.21203/rs.3.rs-4816139/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4816139/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u0026nbsp;\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClinical utilization of machine learning is hampered by the lack of interpretability inherent in most non-linear black box modeling approaches, reducing trust among clinicians and regulators. Advanced large language models, such as GPT-4o, offer a potential framework for integrating medical knowledge into these models, potentially enhancing their interpretability.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u0026nbsp;\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur study utilizes GPT-4o to generate detailed medical feature descriptions, which are aggregated into a comprehensive corpus and processed using TF-IDF vectorization. We then apply fuzzy C-means clustering to these vectorized features to identify significant mortality cause-specific feature clusters. A physician reviews the resulting clusters, validating their relevance to specific mortality causes in mechanically ventilated ICU patients. Subsequently, the resulting clusters inform the creation of weak mortality classifiers, which are combined into a strong classifier using boosting techniques, ultimately producing a GPT-enhanced boosting model for ICU mortality prediction.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis study enrolled 16,018 mechanically ventilated ICU patients, divided into derivation (12,758) and validation (3,260) cohorts, to develop and evaluate a GPT-enhanced boosting model for predicting in-ICU mortality. Leveraging GPT-4o, we implemented an automated process for clustering mortality cause-specific features, resulting in six feature clusters: Liver Failure, Infection, Renal Failure, Hypoxia, Cardiac Failure, and Mechanical Ventilation. This approach significantly improved upon previous manual methods, automating the reconstruction of structured boosting models. While the GPT-enhanced model showed similar predictive accuracy to an XGBoost model, it demonstrated superior interpretability and clinical relevance by incorporating a wider array of features and providing a hierarchical structure of feature importance aligned with medical knowledge.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u0026nbsp;\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe introduce a novel approach to predicting in-ICU mortality for mechanically ventilated patients using a GPT-enhanced boosting model. Our methodology demonstrates the potential of integrating large language models with traditional machine learning techniques to create interpretable and clinically relevant predictive models.\u003c/p\u003e","manuscriptTitle":"GPT-4o and the Quest for Machine Learning Interpretability in ICU Mortality Prediction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-08-06 18:15:48","doi":"10.21203/rs.3.rs-4816139/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-02-28T14:03:47+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-02-24T10:48:12+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"208510925184234101823995201093529591908","date":"2025-02-04T08:52:52+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-01-21T22:01:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"304504606287183388302124343652761357951","date":"2025-01-10T15:04:42+00:00","index":"hide","fulltext":""},{"type":"editorInvited","content":"","date":"2024-12-02T13:53:53+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"275641440595244992816670182776044674396","date":"2024-09-30T07:19:14+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"303660308627683935830306921359377536435","date":"2024-08-18T03:58:14+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-08-15T20:22:55+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-08-01T11:40:24+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-08-01T11:39:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2024-07-28T09:55:06+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5cd24aba-19cf-49c3-b618-aa2de550ccb6","owner":[],"postedDate":"August 6th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-10-20T16:09:25+00:00","versionOfRecord":{"articleIdentity":"rs-4816139","link":"https://doi.org/10.1186/s12911-025-03224-z","journal":{"identity":"bmc-medical-informatics-and-decision-making","isVorOnly":false,"title":"BMC Medical Informatics and Decision Making"},"publishedOn":"2025-10-13 15:58:08","publishedOnDateReadable":"October 13th, 2025"},"versionCreatedAt":"2024-08-06 18:15:48","video":"","vorDoi":"10.1186/s12911-025-03224-z","vorDoiUrl":"https://doi.org/10.1186/s12911-025-03224-z","workflowStages":[]},"version":"v1","identity":"rs-4816139","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4816139","identity":"rs-4816139","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}