LSTM for septic shock: Adding unreliable labels to reliable predictions

Yuan Zhang, Chen Lin, Min Chi, Julie Ivy, Muge Capan, Jeanne M. Huddleston

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Sepsis is a leading cause of death over the world and septic shock, the most severe complication of sepsis, reaches a mortality rate as high as 50%. Early diagnosis and treatment can prevent most morbidity and mortality. Nowadays, the increasing availability of the electronic health records (EHRs) has generated great interests in developing models to predict acute medical conditions such as septic shock. However, septic shock prediction faces two major challenges : 1) how to capture the informative progression of septic shock in a long visit to hospital of a patient; and 2) how to obtain reliable predictions without well-established moment-by-moment ground-truth labels for septic shock. In this work, we proposed a generic framework to predict septic shock based on Long-Short Term Memory (LSTM) model, which is capable of memorizing temporal dependencies over a long period. The framework integrates two levels of imperfect yet informative labels to jointly learn the discriminative patterns of septic shock: ICD-9 code as the visit-level label and the clinical criteria designed by domain experts as the moment-by-moment event-level label. We evaluate our method on a real-world data extracted from an EHR system constituted by 12,954 visits and 1,348,625 events, and compare it against multiple baselines. The robustness of the method is validated using three sets of clinician-proposed adjusted ground-truth labels. Also, we explore whether the framework is effective for the early prediction of the patients developing septic shock. The experimental results demonstrate the superiority of our proposed method in the task of septic shock prediction.

Original languageEnglish (US)
Title of host publicationProceedings - 2017 IEEE International Conference on Big Data, Big Data 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1233-1242
Number of pages10
Volume2018-January
ISBN (Electronic)9781538627143
DOIs
StatePublished - Jan 12 2018
Event5th IEEE International Conference on Big Data, Big Data 2017 - Boston, United States
Duration: Dec 11 2017Dec 14 2017

Other

Other5th IEEE International Conference on Big Data, Big Data 2017
CountryUnited States
CityBoston
Period12/11/1712/14/17

Fingerprint

Memory Term
Labels
Shock
Prediction
Moment
Health
Electronics
Long short-term memory
Availability
Predict
Morbidity
Mortality Rate
Memory Model
Complications
Progression
Mortality
Imperfect
Acute
Baseline
Integrate

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Hardware and Architecture
  • Information Systems
  • Information Systems and Management
  • Control and Optimization

Cite this

Zhang, Y., Lin, C., Chi, M., Ivy, J., Capan, M., & Huddleston, J. M. (2018). LSTM for septic shock: Adding unreliable labels to reliable predictions. In Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017 (Vol. 2018-January, pp. 1233-1242). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/BigData.2017.8258049

LSTM for septic shock : Adding unreliable labels to reliable predictions. / Zhang, Yuan; Lin, Chen; Chi, Min; Ivy, Julie; Capan, Muge; Huddleston, Jeanne M.

Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017. Vol. 2018-January Institute of Electrical and Electronics Engineers Inc., 2018. p. 1233-1242.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Zhang, Y, Lin, C, Chi, M, Ivy, J, Capan, M & Huddleston, JM 2018, LSTM for septic shock: Adding unreliable labels to reliable predictions. in Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017. vol. 2018-January, Institute of Electrical and Electronics Engineers Inc., pp. 1233-1242, 5th IEEE International Conference on Big Data, Big Data 2017, Boston, United States, 12/11/17. https://doi.org/10.1109/BigData.2017.8258049
Zhang Y, Lin C, Chi M, Ivy J, Capan M, Huddleston JM. LSTM for septic shock: Adding unreliable labels to reliable predictions. In Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017. Vol. 2018-January. Institute of Electrical and Electronics Engineers Inc. 2018. p. 1233-1242 https://doi.org/10.1109/BigData.2017.8258049
Zhang, Yuan ; Lin, Chen ; Chi, Min ; Ivy, Julie ; Capan, Muge ; Huddleston, Jeanne M. / LSTM for septic shock : Adding unreliable labels to reliable predictions. Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017. Vol. 2018-January Institute of Electrical and Electronics Engineers Inc., 2018. pp. 1233-1242
@inproceedings{321365cd2c4942978fbb11c93d26bdb7,
title = "LSTM for septic shock: Adding unreliable labels to reliable predictions",
abstract = "Sepsis is a leading cause of death over the world and septic shock, the most severe complication of sepsis, reaches a mortality rate as high as 50{\%}. Early diagnosis and treatment can prevent most morbidity and mortality. Nowadays, the increasing availability of the electronic health records (EHRs) has generated great interests in developing models to predict acute medical conditions such as septic shock. However, septic shock prediction faces two major challenges : 1) how to capture the informative progression of septic shock in a long visit to hospital of a patient; and 2) how to obtain reliable predictions without well-established moment-by-moment ground-truth labels for septic shock. In this work, we proposed a generic framework to predict septic shock based on Long-Short Term Memory (LSTM) model, which is capable of memorizing temporal dependencies over a long period. The framework integrates two levels of imperfect yet informative labels to jointly learn the discriminative patterns of septic shock: ICD-9 code as the visit-level label and the clinical criteria designed by domain experts as the moment-by-moment event-level label. We evaluate our method on a real-world data extracted from an EHR system constituted by 12,954 visits and 1,348,625 events, and compare it against multiple baselines. The robustness of the method is validated using three sets of clinician-proposed adjusted ground-truth labels. Also, we explore whether the framework is effective for the early prediction of the patients developing septic shock. The experimental results demonstrate the superiority of our proposed method in the task of septic shock prediction.",
author = "Yuan Zhang and Chen Lin and Min Chi and Julie Ivy and Muge Capan and Huddleston, {Jeanne M.}",
year = "2018",
month = "1",
day = "12",
doi = "10.1109/BigData.2017.8258049",
language = "English (US)",
volume = "2018-January",
pages = "1233--1242",
booktitle = "Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - LSTM for septic shock

T2 - Adding unreliable labels to reliable predictions

AU - Zhang, Yuan

AU - Lin, Chen

AU - Chi, Min

AU - Ivy, Julie

AU - Capan, Muge

AU - Huddleston, Jeanne M.

PY - 2018/1/12

Y1 - 2018/1/12

N2 - Sepsis is a leading cause of death over the world and septic shock, the most severe complication of sepsis, reaches a mortality rate as high as 50%. Early diagnosis and treatment can prevent most morbidity and mortality. Nowadays, the increasing availability of the electronic health records (EHRs) has generated great interests in developing models to predict acute medical conditions such as septic shock. However, septic shock prediction faces two major challenges : 1) how to capture the informative progression of septic shock in a long visit to hospital of a patient; and 2) how to obtain reliable predictions without well-established moment-by-moment ground-truth labels for septic shock. In this work, we proposed a generic framework to predict septic shock based on Long-Short Term Memory (LSTM) model, which is capable of memorizing temporal dependencies over a long period. The framework integrates two levels of imperfect yet informative labels to jointly learn the discriminative patterns of septic shock: ICD-9 code as the visit-level label and the clinical criteria designed by domain experts as the moment-by-moment event-level label. We evaluate our method on a real-world data extracted from an EHR system constituted by 12,954 visits and 1,348,625 events, and compare it against multiple baselines. The robustness of the method is validated using three sets of clinician-proposed adjusted ground-truth labels. Also, we explore whether the framework is effective for the early prediction of the patients developing septic shock. The experimental results demonstrate the superiority of our proposed method in the task of septic shock prediction.

AB - Sepsis is a leading cause of death over the world and septic shock, the most severe complication of sepsis, reaches a mortality rate as high as 50%. Early diagnosis and treatment can prevent most morbidity and mortality. Nowadays, the increasing availability of the electronic health records (EHRs) has generated great interests in developing models to predict acute medical conditions such as septic shock. However, septic shock prediction faces two major challenges : 1) how to capture the informative progression of septic shock in a long visit to hospital of a patient; and 2) how to obtain reliable predictions without well-established moment-by-moment ground-truth labels for septic shock. In this work, we proposed a generic framework to predict septic shock based on Long-Short Term Memory (LSTM) model, which is capable of memorizing temporal dependencies over a long period. The framework integrates two levels of imperfect yet informative labels to jointly learn the discriminative patterns of septic shock: ICD-9 code as the visit-level label and the clinical criteria designed by domain experts as the moment-by-moment event-level label. We evaluate our method on a real-world data extracted from an EHR system constituted by 12,954 visits and 1,348,625 events, and compare it against multiple baselines. The robustness of the method is validated using three sets of clinician-proposed adjusted ground-truth labels. Also, we explore whether the framework is effective for the early prediction of the patients developing septic shock. The experimental results demonstrate the superiority of our proposed method in the task of septic shock prediction.

UR - http://www.scopus.com/inward/record.url?scp=85047867002&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85047867002&partnerID=8YFLogxK

U2 - 10.1109/BigData.2017.8258049

DO - 10.1109/BigData.2017.8258049

M3 - Conference contribution

AN - SCOPUS:85047867002

VL - 2018-January

SP - 1233

EP - 1242

BT - Proceedings - 2017 IEEE International Conference on Big Data, Big Data 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -