Interpreting patient-Specific risk prediction using contextual decomposition of BiLSTMs: Application to children with asthma

Rawan Alsaad, Qutaibah Malluhi, Ibrahim Janahi, Sabri Boughorbel

Research output: Contribution to journalArticle

Abstract

Background: Predictive modeling with longitudinal electronic health record (EHR) data offers great promise for accelerating personalized medicine and better informs clinical decision-making. Recently, deep learning models have achieved state-of-The-Art performance for many healthcare prediction tasks. However, deep models lack interpretability, which is integral to successful decision-making and can lead to better patient care. In this paper, we build upon the contextual decomposition (CD) method, an algorithm for producing importance scores from long short-Term memory networks (LSTMs). We extend the method to bidirectional LSTMs (BiLSTMs) and use it in the context of predicting future clinical outcomes using patients' EHR historical visits. Methods: We use a real EHR dataset comprising 11071 patients, to evaluate and compare CD interpretations from LSTM and BiLSTM models. First, we train LSTM and BiLSTM models for the task of predicting which pre-school children with respiratory system-related complications will have asthma at school-Age. After that, we conduct quantitative and qualitative analysis to evaluate the CD interpretations produced by the contextual decomposition of the trained models. In addition, we develop an interactive visualization to demonstrate the utility of CD scores in explaining predicted outcomes. Results: Our experimental evaluation demonstrate that whenever a clear visit-level pattern exists, the models learn that pattern and the contextual decomposition can appropriately attribute the prediction to the correct pattern. In addition, the results confirm that the CD scores agree to a large extent with the importance scores generated using logistic regression coefficients. Our main insight was that rather than interpreting the attribution of individual visits to the predicted outcome, we could instead attribute a model's prediction to a group of visits. Conclusion: We presented a quantitative and qualitative evidence that CD interpretations can explain patient-specific predictions using CD attributions of individual visits or a group of visits.

Original languageEnglish
Article number214
JournalBMC Medical Informatics and Decision Making
Volume19
Issue number1
DOIs
Publication statusPublished - 8 Nov 2019

Fingerprint

Long-Term Memory
Short-Term Memory
Electronic Health Records
Asthma
Precision Medicine
Respiratory System
Decision Making
Patient Care
Logistic Models
Learning
Delivery of Health Care

Keywords

  • Deep learning
  • Electronic health record
  • Interpretability
  • Predictive models

ASJC Scopus subject areas

  • Health Policy
  • Health Informatics

Cite this

@article{5720692c031246c8bb67fe7a995f0196,
title = "Interpreting patient-Specific risk prediction using contextual decomposition of BiLSTMs: Application to children with asthma",
abstract = "Background: Predictive modeling with longitudinal electronic health record (EHR) data offers great promise for accelerating personalized medicine and better informs clinical decision-making. Recently, deep learning models have achieved state-of-The-Art performance for many healthcare prediction tasks. However, deep models lack interpretability, which is integral to successful decision-making and can lead to better patient care. In this paper, we build upon the contextual decomposition (CD) method, an algorithm for producing importance scores from long short-Term memory networks (LSTMs). We extend the method to bidirectional LSTMs (BiLSTMs) and use it in the context of predicting future clinical outcomes using patients' EHR historical visits. Methods: We use a real EHR dataset comprising 11071 patients, to evaluate and compare CD interpretations from LSTM and BiLSTM models. First, we train LSTM and BiLSTM models for the task of predicting which pre-school children with respiratory system-related complications will have asthma at school-Age. After that, we conduct quantitative and qualitative analysis to evaluate the CD interpretations produced by the contextual decomposition of the trained models. In addition, we develop an interactive visualization to demonstrate the utility of CD scores in explaining predicted outcomes. Results: Our experimental evaluation demonstrate that whenever a clear visit-level pattern exists, the models learn that pattern and the contextual decomposition can appropriately attribute the prediction to the correct pattern. In addition, the results confirm that the CD scores agree to a large extent with the importance scores generated using logistic regression coefficients. Our main insight was that rather than interpreting the attribution of individual visits to the predicted outcome, we could instead attribute a model's prediction to a group of visits. Conclusion: We presented a quantitative and qualitative evidence that CD interpretations can explain patient-specific predictions using CD attributions of individual visits or a group of visits.",
keywords = "Deep learning, Electronic health record, Interpretability, Predictive models",
author = "Rawan Alsaad and Qutaibah Malluhi and Ibrahim Janahi and Sabri Boughorbel",
year = "2019",
month = "11",
day = "8",
doi = "10.1186/s12911-019-0951-4",
language = "English",
volume = "19",
journal = "BMC Medical Informatics and Decision Making",
issn = "1472-6947",
publisher = "BioMed Central",
number = "1",

}

TY - JOUR

T1 - Interpreting patient-Specific risk prediction using contextual decomposition of BiLSTMs

T2 - Application to children with asthma

AU - Alsaad, Rawan

AU - Malluhi, Qutaibah

AU - Janahi, Ibrahim

AU - Boughorbel, Sabri

PY - 2019/11/8

Y1 - 2019/11/8

N2 - Background: Predictive modeling with longitudinal electronic health record (EHR) data offers great promise for accelerating personalized medicine and better informs clinical decision-making. Recently, deep learning models have achieved state-of-The-Art performance for many healthcare prediction tasks. However, deep models lack interpretability, which is integral to successful decision-making and can lead to better patient care. In this paper, we build upon the contextual decomposition (CD) method, an algorithm for producing importance scores from long short-Term memory networks (LSTMs). We extend the method to bidirectional LSTMs (BiLSTMs) and use it in the context of predicting future clinical outcomes using patients' EHR historical visits. Methods: We use a real EHR dataset comprising 11071 patients, to evaluate and compare CD interpretations from LSTM and BiLSTM models. First, we train LSTM and BiLSTM models for the task of predicting which pre-school children with respiratory system-related complications will have asthma at school-Age. After that, we conduct quantitative and qualitative analysis to evaluate the CD interpretations produced by the contextual decomposition of the trained models. In addition, we develop an interactive visualization to demonstrate the utility of CD scores in explaining predicted outcomes. Results: Our experimental evaluation demonstrate that whenever a clear visit-level pattern exists, the models learn that pattern and the contextual decomposition can appropriately attribute the prediction to the correct pattern. In addition, the results confirm that the CD scores agree to a large extent with the importance scores generated using logistic regression coefficients. Our main insight was that rather than interpreting the attribution of individual visits to the predicted outcome, we could instead attribute a model's prediction to a group of visits. Conclusion: We presented a quantitative and qualitative evidence that CD interpretations can explain patient-specific predictions using CD attributions of individual visits or a group of visits.

AB - Background: Predictive modeling with longitudinal electronic health record (EHR) data offers great promise for accelerating personalized medicine and better informs clinical decision-making. Recently, deep learning models have achieved state-of-The-Art performance for many healthcare prediction tasks. However, deep models lack interpretability, which is integral to successful decision-making and can lead to better patient care. In this paper, we build upon the contextual decomposition (CD) method, an algorithm for producing importance scores from long short-Term memory networks (LSTMs). We extend the method to bidirectional LSTMs (BiLSTMs) and use it in the context of predicting future clinical outcomes using patients' EHR historical visits. Methods: We use a real EHR dataset comprising 11071 patients, to evaluate and compare CD interpretations from LSTM and BiLSTM models. First, we train LSTM and BiLSTM models for the task of predicting which pre-school children with respiratory system-related complications will have asthma at school-Age. After that, we conduct quantitative and qualitative analysis to evaluate the CD interpretations produced by the contextual decomposition of the trained models. In addition, we develop an interactive visualization to demonstrate the utility of CD scores in explaining predicted outcomes. Results: Our experimental evaluation demonstrate that whenever a clear visit-level pattern exists, the models learn that pattern and the contextual decomposition can appropriately attribute the prediction to the correct pattern. In addition, the results confirm that the CD scores agree to a large extent with the importance scores generated using logistic regression coefficients. Our main insight was that rather than interpreting the attribution of individual visits to the predicted outcome, we could instead attribute a model's prediction to a group of visits. Conclusion: We presented a quantitative and qualitative evidence that CD interpretations can explain patient-specific predictions using CD attributions of individual visits or a group of visits.

KW - Deep learning

KW - Electronic health record

KW - Interpretability

KW - Predictive models

UR - http://www.scopus.com/inward/record.url?scp=85074715143&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074715143&partnerID=8YFLogxK

U2 - 10.1186/s12911-019-0951-4

DO - 10.1186/s12911-019-0951-4

M3 - Article

C2 - 31703676

AN - SCOPUS:85074715143

VL - 19

JO - BMC Medical Informatics and Decision Making

JF - BMC Medical Informatics and Decision Making

SN - 1472-6947

IS - 1

M1 - 214

ER -