Healthcare predictive analytics for risk profiling in chronic care: A Bayesian multitask learning approach

Yu Kai Lin, Hsinchun Chen, Randall A. Brown, Shu Hsing Li, Hung Jen Yang

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

Clinical intelligence about a patient's risk of future adverse health events can support clinical decision making in personalized and preventive care. Healthcare predictive analytics using electronic health records offers a promising direction to address the challenging tasks of risk profiling. Patients with chronic diseases often face risks of not just one, but an array of adverse health events. However, existing risk models typically focus on one specific event and do not predict multiple outcomes. To attain enhanced risk profiling, we adopt the design science paradigm and propose a principled approach called Bayesian multitask learning (BMTL). Considering the model development for an event as a single task, our BMTL approach is to coordinate a set of baseline models-one for each event-and communicate training information across the models. The BMTL approach allows healthcare providers to achieve multifaceted risk profiling and model an arbitrary number of events simultaneously. Our experimental evaluations demonstrate that the BMTL approach attains an improved predictive performance when compared with the alternatives that model multiple events separately. We also find that, in most cases, the BMTL approach significantly outperforms existing multitask learning techniques. More importantly, our analysis shows that the BMTL approach can create significant potential impacts on clinical practice in reducing the failures and delays in preventive interventions. We discuss several implications of this study for health IT, big data and predictive analytics, and design science research.

Original languageEnglish (US)
Pages (from-to)473-495
Number of pages23
JournalMIS Quarterly: Management Information Systems
Volume41
Issue number2
StatePublished - Jun 1 2017

Fingerprint

Health
Predictive analytics
Profiling
Healthcare
Decision making
Big data

Keywords

  • Bayesian data analysis
  • Design science
  • Electronic health records
  • Health IT
  • Healthcare predictive analytics
  • Multitask learning

ASJC Scopus subject areas

  • Management Information Systems
  • Information Systems
  • Computer Science Applications
  • Information Systems and Management

Cite this

Healthcare predictive analytics for risk profiling in chronic care : A Bayesian multitask learning approach. / Lin, Yu Kai; Chen, Hsinchun; Brown, Randall A.; Li, Shu Hsing; Yang, Hung Jen.

In: MIS Quarterly: Management Information Systems, Vol. 41, No. 2, 01.06.2017, p. 473-495.

Research output: Contribution to journalArticle

@article{9894b00e88874c70a0f7602116735a6e,
title = "Healthcare predictive analytics for risk profiling in chronic care: A Bayesian multitask learning approach",
abstract = "Clinical intelligence about a patient's risk of future adverse health events can support clinical decision making in personalized and preventive care. Healthcare predictive analytics using electronic health records offers a promising direction to address the challenging tasks of risk profiling. Patients with chronic diseases often face risks of not just one, but an array of adverse health events. However, existing risk models typically focus on one specific event and do not predict multiple outcomes. To attain enhanced risk profiling, we adopt the design science paradigm and propose a principled approach called Bayesian multitask learning (BMTL). Considering the model development for an event as a single task, our BMTL approach is to coordinate a set of baseline models-one for each event-and communicate training information across the models. The BMTL approach allows healthcare providers to achieve multifaceted risk profiling and model an arbitrary number of events simultaneously. Our experimental evaluations demonstrate that the BMTL approach attains an improved predictive performance when compared with the alternatives that model multiple events separately. We also find that, in most cases, the BMTL approach significantly outperforms existing multitask learning techniques. More importantly, our analysis shows that the BMTL approach can create significant potential impacts on clinical practice in reducing the failures and delays in preventive interventions. We discuss several implications of this study for health IT, big data and predictive analytics, and design science research.",
keywords = "Bayesian data analysis, Design science, Electronic health records, Health IT, Healthcare predictive analytics, Multitask learning",
author = "Lin, {Yu Kai} and Hsinchun Chen and Brown, {Randall A.} and Li, {Shu Hsing} and Yang, {Hung Jen}",
year = "2017",
month = "6",
day = "1",
language = "English (US)",
volume = "41",
pages = "473--495",
journal = "MIS Quarterly: Management Information Systems",
issn = "0276-7783",
publisher = "Management Information Systems Research Center",
number = "2",

}

TY - JOUR

T1 - Healthcare predictive analytics for risk profiling in chronic care

T2 - A Bayesian multitask learning approach

AU - Lin, Yu Kai

AU - Chen, Hsinchun

AU - Brown, Randall A.

AU - Li, Shu Hsing

AU - Yang, Hung Jen

PY - 2017/6/1

Y1 - 2017/6/1

N2 - Clinical intelligence about a patient's risk of future adverse health events can support clinical decision making in personalized and preventive care. Healthcare predictive analytics using electronic health records offers a promising direction to address the challenging tasks of risk profiling. Patients with chronic diseases often face risks of not just one, but an array of adverse health events. However, existing risk models typically focus on one specific event and do not predict multiple outcomes. To attain enhanced risk profiling, we adopt the design science paradigm and propose a principled approach called Bayesian multitask learning (BMTL). Considering the model development for an event as a single task, our BMTL approach is to coordinate a set of baseline models-one for each event-and communicate training information across the models. The BMTL approach allows healthcare providers to achieve multifaceted risk profiling and model an arbitrary number of events simultaneously. Our experimental evaluations demonstrate that the BMTL approach attains an improved predictive performance when compared with the alternatives that model multiple events separately. We also find that, in most cases, the BMTL approach significantly outperforms existing multitask learning techniques. More importantly, our analysis shows that the BMTL approach can create significant potential impacts on clinical practice in reducing the failures and delays in preventive interventions. We discuss several implications of this study for health IT, big data and predictive analytics, and design science research.

AB - Clinical intelligence about a patient's risk of future adverse health events can support clinical decision making in personalized and preventive care. Healthcare predictive analytics using electronic health records offers a promising direction to address the challenging tasks of risk profiling. Patients with chronic diseases often face risks of not just one, but an array of adverse health events. However, existing risk models typically focus on one specific event and do not predict multiple outcomes. To attain enhanced risk profiling, we adopt the design science paradigm and propose a principled approach called Bayesian multitask learning (BMTL). Considering the model development for an event as a single task, our BMTL approach is to coordinate a set of baseline models-one for each event-and communicate training information across the models. The BMTL approach allows healthcare providers to achieve multifaceted risk profiling and model an arbitrary number of events simultaneously. Our experimental evaluations demonstrate that the BMTL approach attains an improved predictive performance when compared with the alternatives that model multiple events separately. We also find that, in most cases, the BMTL approach significantly outperforms existing multitask learning techniques. More importantly, our analysis shows that the BMTL approach can create significant potential impacts on clinical practice in reducing the failures and delays in preventive interventions. We discuss several implications of this study for health IT, big data and predictive analytics, and design science research.

KW - Bayesian data analysis

KW - Design science

KW - Electronic health records

KW - Health IT

KW - Healthcare predictive analytics

KW - Multitask learning

UR - http://www.scopus.com/inward/record.url?scp=85019843412&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85019843412&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:85019843412

VL - 41

SP - 473

EP - 495

JO - MIS Quarterly: Management Information Systems

JF - MIS Quarterly: Management Information Systems

SN - 0276-7783

IS - 2

ER -