Scaling convolutional neural networks achieves expert-level seizure detection in neonatal EEG

Scaling convolutional neural networks achieves expert-level seizure detection in neonatal EEG | 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 Scaling convolutional neural networks achieves expert-level seizure detection in neonatal EEG Robert Hogan, Sean Mathieson, Aurel Luca, Soraia Ventura, Sean Griffin, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4682370/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 08 Jan, 2025 Read the published version in npj Digital Medicine → Version 1 posted 12 You are reading this latest preprint version Abstract Neonatal seizures require urgent treatment but can go undetected without expert EEG monitoring. We develop and validate a seizure-detection model using retrospective EEG data from 332 neonates. A convolutional neural network was developed on over 50k hours (n=202) of annotated single-channel EEG containing 12k seizure events. This model was validated on two independent multi-reviewer datasets (n=51 and n=79). Increasing data and model size improves performance: Matthews correlation coefficient (MCC) and Pearson's correlation (r) increased by up to 50% (15%) with data (model) scaling. The largest model (21m parameters) achieves state-of-the-art on an open-access dataset (MCC=0.764, r=0.824, and AUC=0.982). This model also attains expert-level performance on both validation sets, a first in this field, with no significant difference in inter-rater agreement when the model replaces an expert (Δκ 0.05). Health sciences/Neurology/Neurological disorders/Encephalopathy Health sciences/Health care/Diagnosis Health sciences/Biomarkers/Diagnostic markers Health sciences/Neurology/Neurological disorders/Epilepsy Health sciences/Neurology/Neurological disorders/Hypoxic–ischaemic encephalopathy Full Text Additional Declarations There is a conflict of interest All authors are affiliated with Cergenx Ltd, a company developing neuromonitoring technologies for newborns: GB and SG are co-founders; RH, AL, and JOT are employees; and SM and SV were paid contractors Cite Share Download PDF Status: Published Journal Publication published 08 Jan, 2025 Read the published version in npj Digital Medicine → Version 1 posted Editorial decision: revise 02 Sep, 2024 Review # 3 received at journal 31 Aug, 2024 Review # 2 received at journal 23 Aug, 2024 Reviewer # 3 agreed at journal 14 Aug, 2024 Review # 1 received at journal 13 Aug, 2024 Reviewer # 2 agreed at journal 05 Aug, 2024 Reviewer # 1 agreed at journal 05 Aug, 2024 Reviewers invited by journal 04 Aug, 2024 Editor assigned by journal 15 Jul, 2024 Submission checks completed at journal 15 Jul, 2024 First submitted to journal 12 Jul, 2024 Unknown event 11 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-4682370","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":335793694,"identity":"5827c793-fefa-4ac0-85dd-70ff43558050","order_by":0,"name":"Robert Hogan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYDACCTBpQ7qWNNK1HCZBh/ns5sMvfu44L88vffzi48I2Bnl+sQP4tcjcOZZm2XvmtuHMvpxi4xlnGAxnzk4g4C6JHDMD3rbbCQZneNKkeSoYEgxuE6HF8G/buQT7Mzzpv3kMiNNi/Ji37UCCAQ/7MWYibUlLY5ZtSzaccYaHWZrnjAQxfkk+/PFtm508fw/7w8+8bTbAoCOgBQjYIFHDAPQILJoIAeYPEJr9AVHKR8EoGAWjYOQBAOZoPHc6hsyHAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0009-0006-9791-1581","institution":"CergenX Ltd","correspondingAuthor":true,"prefix":"","firstName":"Robert","middleName":"","lastName":"Hogan","suffix":""},{"id":335793695,"identity":"cf0062f9-62e0-48ad-a6cc-d45a8f5f9920","order_by":1,"name":"Sean Mathieson","email":"","orcid":"","institution":"CergenX Ltd","correspondingAuthor":false,"prefix":"","firstName":"Sean","middleName":"","lastName":"Mathieson","suffix":""},{"id":335793696,"identity":"d5435617-7900-4a3e-9d47-e49cfed936b2","order_by":2,"name":"Aurel Luca","email":"","orcid":"","institution":"CergenX Ltd","correspondingAuthor":false,"prefix":"","firstName":"Aurel","middleName":"","lastName":"Luca","suffix":""},{"id":335793697,"identity":"922e1504-b817-4315-b316-7e851915306c","order_by":3,"name":"Soraia Ventura","email":"","orcid":"https://orcid.org/0000-0003-2662-1924","institution":"CergenX Ltd","correspondingAuthor":false,"prefix":"","firstName":"Soraia","middleName":"","lastName":"Ventura","suffix":""},{"id":335793698,"identity":"89b00dd7-b0e2-4309-878c-d01ec2989165","order_by":4,"name":"Sean Griffin","email":"","orcid":"","institution":"CergenX Ltd","correspondingAuthor":false,"prefix":"","firstName":"Sean","middleName":"","lastName":"Griffin","suffix":""},{"id":335793699,"identity":"9ed73c37-1d84-40e7-a49a-634f7f3533e3","order_by":5,"name":"Geraldine Boylan","email":"","orcid":"","institution":"CergenX Ltd","correspondingAuthor":false,"prefix":"","firstName":"Geraldine","middleName":"","lastName":"Boylan","suffix":""},{"id":335793700,"identity":"f70c18d0-b445-490a-a03c-6e431d538d50","order_by":6,"name":"John O'Toole","email":"","orcid":"https://orcid.org/0000-0001-9021-1183","institution":"CergenX Ltd","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"O'Toole","suffix":""}],"badges":[],"createdAt":"2024-07-03 19:10:28","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4682370/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4682370/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41746-024-01416-x","type":"published","date":"2025-01-08T05:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":73357103,"identity":"0e505367-e40a-42f0-ac9c-d6c96848f524","added_by":"auto","created_at":"2025-01-09 08:12:41","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1753298,"visible":true,"origin":"","legend":"","description":"","filename":"main.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4682370/v1_covered_587d5750-4cf9-4ab3-9351-23d63d56e4dd.pdf"}],"financialInterests":"There is a conflict of interest\nAll authors are affiliated with Cergenx Ltd, a company developing neuromonitoring\r\ntechnologies for newborns: GB and SG are co-founders; RH, AL, and JOT are\r\nemployees; and SM and SV were paid contractors","formattedTitle":"Scaling convolutional neural networks achieves expert-level seizure detection in neonatal EEG","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":"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":"","lastPublishedDoi":"10.21203/rs.3.rs-4682370/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4682370/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Neonatal seizures require urgent treatment but can go undetected without expert EEG monitoring. We develop and validate a seizure-detection model using retrospective EEG data from 332 neonates. A convolutional neural network was developed on over 50k hours (n=202) of annotated single-channel EEG containing 12k seizure events. This model was validated on two independent multi-reviewer datasets (n=51 and n=79).\r\n\r\nIncreasing data and model size improves performance: Matthews correlation coefficient (MCC) and Pearson's correlation (r) increased by up to 50% (15%) with data (model) scaling. The largest model (21m parameters) achieves state-of-the-art on an open-access dataset (MCC=0.764, r=0.824, and AUC=0.982). This model also attains expert-level performance on both validation sets, a first in this field, with no significant difference in inter-rater agreement when the model replaces an expert (Δκ \u003c -0.095, p\u003e0.05).","manuscriptTitle":"Scaling convolutional neural networks achieves expert-level seizure detection in neonatal EEG","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-08-29 18:45:36","doi":"10.21203/rs.3.rs-4682370/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"revise","date":"2024-09-02T12:09:50+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"This content is not available.","date":"2024-08-31T11:07:02+00:00","index":3,"fulltext":"This content is not available."},{"type":"editorInvitedReview","content":"This content is not available.","date":"2024-08-23T09:06:35+00:00","index":2,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2024-08-14T10:36:48+00:00","index":3,"fulltext":"This content is not available."},{"type":"editorInvitedReview","content":"This content is not available.","date":"2024-08-13T12:45:33+00:00","index":1,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2024-08-05T10:12:52+00:00","index":2,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2024-08-05T07:43:57+00:00","index":1,"fulltext":"This content is not available."},{"type":"reviewersInvited","content":"","date":"2024-08-04T16:32:52+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-07-15T15:34:07+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-07-15T07:10:59+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Digital Medicine","date":"2024-07-12T22:00:01+00:00","index":"","fulltext":""},{"type":"checksFailed","content":"","date":"2024-07-11T05:13:46+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":"041d7395-dd23-4386-a2c7-92ef10d487d4","owner":[],"postedDate":"August 29th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":35562454,"name":"Health sciences/Neurology/Neurological disorders/Encephalopathy"},{"id":35562455,"name":"Health sciences/Health care/Diagnosis"},{"id":35562456,"name":"Health sciences/Biomarkers/Diagnostic markers"},{"id":35562457,"name":"Health sciences/Neurology/Neurological disorders/Epilepsy"},{"id":35562458,"name":"Health sciences/Neurology/Neurological disorders/Hypoxic\u0026#x2013;ischaemic encephalopathy"}],"tags":[],"updatedAt":"2025-01-09T08:12:35+00:00","versionOfRecord":{"articleIdentity":"rs-4682370","link":"https://doi.org/10.1038/s41746-024-01416-x","journal":{"identity":"npj-digital-medicine","isVorOnly":false,"title":"npj Digital Medicine"},"publishedOn":"2025-01-08 05:00:00","publishedOnDateReadable":"January 8th, 2025"},"versionCreatedAt":"2024-08-29 18:45:36","video":"","vorDoi":"10.1038/s41746-024-01416-x","vorDoiUrl":"https://doi.org/10.1038/s41746-024-01416-x","workflowStages":[]},"version":"v1","identity":"rs-4682370","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4682370","identity":"rs-4682370","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}