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

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

Standard

Early detection of sepsis utilizing deep learning on electronic health record event sequences. / Lauritsen, Simon Meyer; Kalør, Mads Ellersgaard; Kongsgaard, Emil Lund et al.

I: Artificial Intelligence in Medicine, Bind 104, 101820, 04.2020.

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

Harvard

Lauritsen, SM, Kalør, ME, Kongsgaard, EL, Lauritsen, KM, Jørgensen, MJ, Lange, J & Thiesson, B 2020, 'Early detection of sepsis utilizing deep learning on electronic health record event sequences', Artificial Intelligence in Medicine, bind 104, 101820. https://doi.org/10.1016/j.artmed.2020.101820

APA

Lauritsen, S. M., Kalør, M. E., Kongsgaard, E. L., Lauritsen, K. M., Jørgensen, M. J., Lange, J., & Thiesson, B. (2020). Early detection of sepsis utilizing deep learning on electronic health record event sequences. Artificial Intelligence in Medicine, 104, [101820]. https://doi.org/10.1016/j.artmed.2020.101820

CBE

Lauritsen SM, Kalør ME, Kongsgaard EL, Lauritsen KM, Jørgensen MJ, Lange J, Thiesson B. 2020. Early detection of sepsis utilizing deep learning on electronic health record event sequences. Artificial Intelligence in Medicine. 104:Article 101820. https://doi.org/10.1016/j.artmed.2020.101820

MLA

Vancouver

Lauritsen SM, Kalør ME, Kongsgaard EL, Lauritsen KM, Jørgensen MJ, Lange J et al. Early detection of sepsis utilizing deep learning on electronic health record event sequences. Artificial Intelligence in Medicine. 2020 apr.;104. 101820. https://doi.org/10.1016/j.artmed.2020.101820

Author

Lauritsen, Simon Meyer ; Kalør, Mads Ellersgaard ; Kongsgaard, Emil Lund et al. / Early detection of sepsis utilizing deep learning on electronic health record event sequences. I: Artificial Intelligence in Medicine. 2020 ; Bind 104.

Bibtex

@article{997c91f839094deb8c26f6c3a95ebd85,
title = "Early detection of sepsis utilizing deep learning on electronic health record event sequences",
abstract = "Background: The timeliness of detection of a sepsis incidence 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) Model evaluations neglect the clinical consequences of a decision to start, or not start, an intervention for sepsis. (2) Models are evaluated shortly before sepsis onset without considering interventions already initiated. (3) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. (4) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hard-coded into the model. Methods: 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 assess model quality by standard concepts of accuracy as well as clinical usefulness, and we suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results: Results show performance ranging from AUROC 0.856 (3 h before sepsis onset) to AUROC 0.756 (24 h before sepsis onset). Evaluating the clinical utility of the model, we find that a large proportion of septic patients did not receive antibiotic treatment or blood culture at the time of the sepsis prediction, and the model could, therefore, facilitate such interventions at an earlier point in time. Conclusion: We present a deep learning system for early detection of sepsis that can learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work. Our system outperforms baseline models, such as gradient boosting, which rely on specific data elements and therefore suffer from many missing values in our dataset.",
keywords = "Clinical decision support systems, Early diagnosis, Electronic health records, Machine learning, Medical informatics, Sepsis",
author = "Lauritsen, {Simon Meyer} and Kal{\o}r, {Mads Ellersgaard} and Kongsgaard, {Emil Lund} and Lauritsen, {Katrine Meyer} and J{\o}rgensen, {Marianne Johansson} and Jeppe Lange and Bo Thiesson",
year = "2020",
month = apr,
doi = "10.1016/j.artmed.2020.101820",
language = "English",
volume = "104",
journal = "Artificial Intelligence in Medicine",
issn = "0933-3657",
publisher = "Elsevier BV",

}

RIS

TY - JOUR

T1 - Early detection of sepsis utilizing deep learning on electronic health record event sequences

AU - Lauritsen, Simon Meyer

AU - Kalør, Mads Ellersgaard

AU - Kongsgaard, Emil Lund

AU - Lauritsen, Katrine Meyer

AU - Jørgensen, Marianne Johansson

AU - Lange, Jeppe

AU - Thiesson, Bo

PY - 2020/4

Y1 - 2020/4

N2 - Background: The timeliness of detection of a sepsis incidence 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) Model evaluations neglect the clinical consequences of a decision to start, or not start, an intervention for sepsis. (2) Models are evaluated shortly before sepsis onset without considering interventions already initiated. (3) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. (4) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hard-coded into the model. Methods: 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 assess model quality by standard concepts of accuracy as well as clinical usefulness, and we suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results: Results show performance ranging from AUROC 0.856 (3 h before sepsis onset) to AUROC 0.756 (24 h before sepsis onset). Evaluating the clinical utility of the model, we find that a large proportion of septic patients did not receive antibiotic treatment or blood culture at the time of the sepsis prediction, and the model could, therefore, facilitate such interventions at an earlier point in time. Conclusion: We present a deep learning system for early detection of sepsis that can learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work. Our system outperforms baseline models, such as gradient boosting, which rely on specific data elements and therefore suffer from many missing values in our dataset.

AB - Background: The timeliness of detection of a sepsis incidence 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) Model evaluations neglect the clinical consequences of a decision to start, or not start, an intervention for sepsis. (2) Models are evaluated shortly before sepsis onset without considering interventions already initiated. (3) Machine learning models are built on a restricted set of clinical parameters, which are not necessarily measured in all departments. (4) Model performance is limited by current knowledge of sepsis, as feature interactions and time dependencies are hard-coded into the model. Methods: 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 assess model quality by standard concepts of accuracy as well as clinical usefulness, and we suggest a retrospective assessment of interventions by looking at intravenous antibiotics and blood cultures preceding the prediction time. Results: Results show performance ranging from AUROC 0.856 (3 h before sepsis onset) to AUROC 0.756 (24 h before sepsis onset). Evaluating the clinical utility of the model, we find that a large proportion of septic patients did not receive antibiotic treatment or blood culture at the time of the sepsis prediction, and the model could, therefore, facilitate such interventions at an earlier point in time. Conclusion: We present a deep learning system for early detection of sepsis that can learn characteristics of the key factors and interactions from the raw event sequence data itself, without relying on a labor-intensive feature extraction work. Our system outperforms baseline models, such as gradient boosting, which rely on specific data elements and therefore suffer from many missing values in our dataset.

KW - Clinical decision support systems

KW - Early diagnosis

KW - Electronic health records

KW - Machine learning

KW - Medical informatics

KW - Sepsis

UR - http://www.scopus.com/inward/record.url?scp=85082865174&partnerID=8YFLogxK

U2 - 10.1016/j.artmed.2020.101820

DO - 10.1016/j.artmed.2020.101820

M3 - Journal article

C2 - 32498999

AN - SCOPUS:85082865174

VL - 104

JO - Artificial Intelligence in Medicine

JF - Artificial Intelligence in Medicine

SN - 0933-3657

M1 - 101820

ER -