Early detection of sepsis utilizing deep learning on electronic health record event sequences

The timeliness of detection of a sepsis event in progress is a crucial factor in the outcome for the patient. Machine learning models built from data in electronic health records can be used as an effective tool for improving this timeliness, but so far the potential for clinical implementations has...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lauritsen, Simon Meyer, Kalør, Mads Ellersgaard, Kongsgaard, Emil Lund, Lauritsen, Katrine Meyer, Jørgensen, Marianne Johansson, Lange, Jeppe, Thiesson, Bo
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Lauritsen, Simon Meyer
Kalør, Mads Ellersgaard
Kongsgaard, Emil Lund
Lauritsen, Katrine Meyer
Jørgensen, Marianne Johansson
Lange, Jeppe
Thiesson, Bo
description The timeliness of detection of a sepsis event in progress is a crucial factor in the outcome for the patient. Machine learning models built from data in electronic health records can be used as an effective tool for improving this timeliness, but so far the potential for clinical implementations has been largely limited to studies in intensive care units. This study will employ a richer data set that will expand the applicability of these models beyond intensive care units. Furthermore, we will circumvent several important limitations that have been found in the literature: 1) Models are evaluated shortly before sepsis onset without considering interventions already initiated. 2) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. 3) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hardcoded into the model. In this study, we present a model to overcome these shortcomings using a deep learning approach on a diverse multicenter data set. We used retrospective data from multiple Danish hospitals over a seven-year period. Our sepsis detection system is constructed as a combination of a convolutional neural network and a long short-term memory network. We suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results show performance ranging from AUROC 0.856 (3 hours before sepsis onset) to AUROC 0.756 (24 hours before sepsis onset). We present a deep learning system for early detection of sepsis that is able to learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work.
doi_str_mv 10.48550/arxiv.1906.02956
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1906_02956</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1906_02956</sourcerecordid><originalsourceid>FETCH-LOGICAL-a676-a008c3f0bc4785267cc59e7521242f63184b86d61a9510878e71accce607652a3</originalsourceid><addsrcrecordid>eNotj8tOwzAURL1hgQofwAr_QILtxI8sUVUoUiU23aLo9uaGWjJOsNOK8vVNC6vRaI5GOow9SFHWTmvxBOnHH0vZCFMK1Whzyz5WkMKJdzQRTn6IfOh5pjH7zA-TD_7Xx895pZEHghQvbYYozHQaoke-JwjTnifCIXWcjhSn-eD7QBEp37GbHkKm-_9csO3LartcF5v317fl86YAY00BQjiserHD2jqtjEXUDVmtpKpVbyrp6p0znZHQaCmcdWQlICIZYY1WUC3Y49_t1a8dk_-CdGovnu3VszoDNdBOYw</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Early detection of sepsis utilizing deep learning on electronic health record event sequences</title><source>arXiv.org</source><creator>Lauritsen, Simon Meyer ; Kalør, Mads Ellersgaard ; Kongsgaard, Emil Lund ; Lauritsen, Katrine Meyer ; Jørgensen, Marianne Johansson ; Lange, Jeppe ; Thiesson, Bo</creator><creatorcontrib>Lauritsen, Simon Meyer ; Kalør, Mads Ellersgaard ; Kongsgaard, Emil Lund ; Lauritsen, Katrine Meyer ; Jørgensen, Marianne Johansson ; Lange, Jeppe ; Thiesson, Bo</creatorcontrib><description>The timeliness of detection of a sepsis event in progress is a crucial factor in the outcome for the patient. Machine learning models built from data in electronic health records can be used as an effective tool for improving this timeliness, but so far the potential for clinical implementations has been largely limited to studies in intensive care units. This study will employ a richer data set that will expand the applicability of these models beyond intensive care units. Furthermore, we will circumvent several important limitations that have been found in the literature: 1) Models are evaluated shortly before sepsis onset without considering interventions already initiated. 2) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. 3) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hardcoded into the model. In this study, we present a model to overcome these shortcomings using a deep learning approach on a diverse multicenter data set. We used retrospective data from multiple Danish hospitals over a seven-year period. Our sepsis detection system is constructed as a combination of a convolutional neural network and a long short-term memory network. We suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results show performance ranging from AUROC 0.856 (3 hours before sepsis onset) to AUROC 0.756 (24 hours before sepsis onset). We present a deep learning system for early detection of sepsis that is able to learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work.</description><identifier>DOI: 10.48550/arxiv.1906.02956</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Applications ; Statistics - Machine Learning</subject><creationdate>2019-06</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1906.02956$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1906.02956$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Lauritsen, Simon Meyer</creatorcontrib><creatorcontrib>Kalør, Mads Ellersgaard</creatorcontrib><creatorcontrib>Kongsgaard, Emil Lund</creatorcontrib><creatorcontrib>Lauritsen, Katrine Meyer</creatorcontrib><creatorcontrib>Jørgensen, Marianne Johansson</creatorcontrib><creatorcontrib>Lange, Jeppe</creatorcontrib><creatorcontrib>Thiesson, Bo</creatorcontrib><title>Early detection of sepsis utilizing deep learning on electronic health record event sequences</title><description>The timeliness of detection of a sepsis event in progress is a crucial factor in the outcome for the patient. Machine learning models built from data in electronic health records can be used as an effective tool for improving this timeliness, but so far the potential for clinical implementations has been largely limited to studies in intensive care units. This study will employ a richer data set that will expand the applicability of these models beyond intensive care units. Furthermore, we will circumvent several important limitations that have been found in the literature: 1) Models are evaluated shortly before sepsis onset without considering interventions already initiated. 2) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. 3) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hardcoded into the model. In this study, we present a model to overcome these shortcomings using a deep learning approach on a diverse multicenter data set. We used retrospective data from multiple Danish hospitals over a seven-year period. Our sepsis detection system is constructed as a combination of a convolutional neural network and a long short-term memory network. We suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results show performance ranging from AUROC 0.856 (3 hours before sepsis onset) to AUROC 0.756 (24 hours before sepsis onset). We present a deep learning system for early detection of sepsis that is able to learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work.</description><subject>Computer Science - Learning</subject><subject>Statistics - Applications</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj8tOwzAURL1hgQofwAr_QILtxI8sUVUoUiU23aLo9uaGWjJOsNOK8vVNC6vRaI5GOow9SFHWTmvxBOnHH0vZCFMK1Whzyz5WkMKJdzQRTn6IfOh5pjH7zA-TD_7Xx895pZEHghQvbYYozHQaoke-JwjTnifCIXWcjhSn-eD7QBEp37GbHkKm-_9csO3LartcF5v317fl86YAY00BQjiserHD2jqtjEXUDVmtpKpVbyrp6p0znZHQaCmcdWQlICIZYY1WUC3Y49_t1a8dk_-CdGovnu3VszoDNdBOYw</recordid><startdate>20190607</startdate><enddate>20190607</enddate><creator>Lauritsen, Simon Meyer</creator><creator>Kalør, Mads Ellersgaard</creator><creator>Kongsgaard, Emil Lund</creator><creator>Lauritsen, Katrine Meyer</creator><creator>Jørgensen, Marianne Johansson</creator><creator>Lange, Jeppe</creator><creator>Thiesson, Bo</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20190607</creationdate><title>Early detection of sepsis utilizing deep learning on electronic health record event sequences</title><author>Lauritsen, Simon Meyer ; Kalør, Mads Ellersgaard ; Kongsgaard, Emil Lund ; Lauritsen, Katrine Meyer ; Jørgensen, Marianne Johansson ; Lange, Jeppe ; Thiesson, Bo</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a676-a008c3f0bc4785267cc59e7521242f63184b86d61a9510878e71accce607652a3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Applications</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Lauritsen, Simon Meyer</creatorcontrib><creatorcontrib>Kalør, Mads Ellersgaard</creatorcontrib><creatorcontrib>Kongsgaard, Emil Lund</creatorcontrib><creatorcontrib>Lauritsen, Katrine Meyer</creatorcontrib><creatorcontrib>Jørgensen, Marianne Johansson</creatorcontrib><creatorcontrib>Lange, Jeppe</creatorcontrib><creatorcontrib>Thiesson, Bo</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Lauritsen, Simon Meyer</au><au>Kalør, Mads Ellersgaard</au><au>Kongsgaard, Emil Lund</au><au>Lauritsen, Katrine Meyer</au><au>Jørgensen, Marianne Johansson</au><au>Lange, Jeppe</au><au>Thiesson, Bo</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Early detection of sepsis utilizing deep learning on electronic health record event sequences</atitle><date>2019-06-07</date><risdate>2019</risdate><abstract>The timeliness of detection of a sepsis event in progress is a crucial factor in the outcome for the patient. Machine learning models built from data in electronic health records can be used as an effective tool for improving this timeliness, but so far the potential for clinical implementations has been largely limited to studies in intensive care units. This study will employ a richer data set that will expand the applicability of these models beyond intensive care units. Furthermore, we will circumvent several important limitations that have been found in the literature: 1) Models are evaluated shortly before sepsis onset without considering interventions already initiated. 2) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. 3) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hardcoded into the model. In this study, we present a model to overcome these shortcomings using a deep learning approach on a diverse multicenter data set. We used retrospective data from multiple Danish hospitals over a seven-year period. Our sepsis detection system is constructed as a combination of a convolutional neural network and a long short-term memory network. We suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results show performance ranging from AUROC 0.856 (3 hours before sepsis onset) to AUROC 0.756 (24 hours before sepsis onset). We present a deep learning system for early detection of sepsis that is able to learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work.</abstract><doi>10.48550/arxiv.1906.02956</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.1906.02956
ispartof
issn
language eng
recordid cdi_arxiv_primary_1906_02956
source arXiv.org
subjects Computer Science - Learning
Statistics - Applications
Statistics - Machine Learning
title Early detection of sepsis utilizing deep learning on electronic health record event sequences
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T15%3A23%3A33IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Early%20detection%20of%20sepsis%20utilizing%20deep%20learning%20on%20electronic%20health%20record%20event%20sequences&rft.au=Lauritsen,%20Simon%20Meyer&rft.date=2019-06-07&rft_id=info:doi/10.48550/arxiv.1906.02956&rft_dat=%3Carxiv_GOX%3E1906_02956%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true