Modeling asynchronous event sequences with RNNs

Stephen Wu, Sijia Liu, Sunghwan Sohn, Sungrim Moon, Chung il Wi, Young J Juhn, Hongfang D Liu

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Sequences of events have often been modeled with computational techniques, but typical preprocessing steps and problem settings do not explicitly address the ramifications of timestamped events. Clinical data, such as is found in electronic health records (EHRs), typically comes with timestamp information. In this work, we define event sequences and their properties: synchronicity, evenness, and co-cardinality; we then show how asynchronous, uneven, and multi-cardinal problem settings can support explicit accountings of relative time. Our evaluation uses the temporally sensitive clinical use case of pediatric asthma, which is a chronic disease with symptoms (and lack thereof) evolving over time. We show several approaches to explicitly incorporating relative time into a recurrent neural network (RNN) model that improve the overall classification of patients into those with no asthma, those with persistent asthma, those in long-term remission, and those who have experienced relapse. We also compare and contrast these results with those in an inpatient intensive care setting.

Original languageEnglish (US)
Pages (from-to)167-177
Number of pages11
JournalJournal of Biomedical Informatics
Volume83
DOIs
StatePublished - Jul 1 2018

Fingerprint

Pediatrics
Recurrent neural networks
Asthma
Health
Neural Networks (Computer)
Electronic Health Records
Critical Care
Inpatients
Chronic Disease
Recurrence

Keywords

  • Asthma
  • Deep learning
  • Electronic health records
  • Temporal data

ASJC Scopus subject areas

  • Computer Science Applications
  • Health Informatics

Cite this

Modeling asynchronous event sequences with RNNs. / Wu, Stephen; Liu, Sijia; Sohn, Sunghwan; Moon, Sungrim; Wi, Chung il; Juhn, Young J; Liu, Hongfang D.

In: Journal of Biomedical Informatics, Vol. 83, 01.07.2018, p. 167-177.

Research output: Contribution to journalArticle

Wu, Stephen ; Liu, Sijia ; Sohn, Sunghwan ; Moon, Sungrim ; Wi, Chung il ; Juhn, Young J ; Liu, Hongfang D. / Modeling asynchronous event sequences with RNNs. In: Journal of Biomedical Informatics. 2018 ; Vol. 83. pp. 167-177.
@article{d954f885f35e4b92b56f1735a6142ba8,
title = "Modeling asynchronous event sequences with RNNs",
abstract = "Sequences of events have often been modeled with computational techniques, but typical preprocessing steps and problem settings do not explicitly address the ramifications of timestamped events. Clinical data, such as is found in electronic health records (EHRs), typically comes with timestamp information. In this work, we define event sequences and their properties: synchronicity, evenness, and co-cardinality; we then show how asynchronous, uneven, and multi-cardinal problem settings can support explicit accountings of relative time. Our evaluation uses the temporally sensitive clinical use case of pediatric asthma, which is a chronic disease with symptoms (and lack thereof) evolving over time. We show several approaches to explicitly incorporating relative time into a recurrent neural network (RNN) model that improve the overall classification of patients into those with no asthma, those with persistent asthma, those in long-term remission, and those who have experienced relapse. We also compare and contrast these results with those in an inpatient intensive care setting.",
keywords = "Asthma, Deep learning, Electronic health records, Temporal data",
author = "Stephen Wu and Sijia Liu and Sunghwan Sohn and Sungrim Moon and Wi, {Chung il} and Juhn, {Young J} and Liu, {Hongfang D}",
year = "2018",
month = "7",
day = "1",
doi = "10.1016/j.jbi.2018.05.016",
language = "English (US)",
volume = "83",
pages = "167--177",
journal = "Journal of Biomedical Informatics",
issn = "1532-0464",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Modeling asynchronous event sequences with RNNs

AU - Wu, Stephen

AU - Liu, Sijia

AU - Sohn, Sunghwan

AU - Moon, Sungrim

AU - Wi, Chung il

AU - Juhn, Young J

AU - Liu, Hongfang D

PY - 2018/7/1

Y1 - 2018/7/1

N2 - Sequences of events have often been modeled with computational techniques, but typical preprocessing steps and problem settings do not explicitly address the ramifications of timestamped events. Clinical data, such as is found in electronic health records (EHRs), typically comes with timestamp information. In this work, we define event sequences and their properties: synchronicity, evenness, and co-cardinality; we then show how asynchronous, uneven, and multi-cardinal problem settings can support explicit accountings of relative time. Our evaluation uses the temporally sensitive clinical use case of pediatric asthma, which is a chronic disease with symptoms (and lack thereof) evolving over time. We show several approaches to explicitly incorporating relative time into a recurrent neural network (RNN) model that improve the overall classification of patients into those with no asthma, those with persistent asthma, those in long-term remission, and those who have experienced relapse. We also compare and contrast these results with those in an inpatient intensive care setting.

AB - Sequences of events have often been modeled with computational techniques, but typical preprocessing steps and problem settings do not explicitly address the ramifications of timestamped events. Clinical data, such as is found in electronic health records (EHRs), typically comes with timestamp information. In this work, we define event sequences and their properties: synchronicity, evenness, and co-cardinality; we then show how asynchronous, uneven, and multi-cardinal problem settings can support explicit accountings of relative time. Our evaluation uses the temporally sensitive clinical use case of pediatric asthma, which is a chronic disease with symptoms (and lack thereof) evolving over time. We show several approaches to explicitly incorporating relative time into a recurrent neural network (RNN) model that improve the overall classification of patients into those with no asthma, those with persistent asthma, those in long-term remission, and those who have experienced relapse. We also compare and contrast these results with those in an inpatient intensive care setting.

KW - Asthma

KW - Deep learning

KW - Electronic health records

KW - Temporal data

UR - http://www.scopus.com/inward/record.url?scp=85048712908&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048712908&partnerID=8YFLogxK

U2 - 10.1016/j.jbi.2018.05.016

DO - 10.1016/j.jbi.2018.05.016

M3 - Article

VL - 83

SP - 167

EP - 177

JO - Journal of Biomedical Informatics

JF - Journal of Biomedical Informatics

SN - 1532-0464

ER -