Bayesian hierarchical vector autoregressive models for patient-level predictive modeling

Feihan Lu, Yao Zheng, Hobart H. Cleveland, III, Chris Burton, David Madigan

Research output: Contribution to journalArticle

Abstract

Predicting health outcomes from longitudinal health histories is of central importance to healthcare. Observational healthcare databases such as patient diary databases provide a rich resource for patient-level predictive modeling. In this paper, we propose a Bayesian hierarchical vector autoregressive (VAR) model to predict medical and psychological conditions using multivariate time series data. Compared to the existing patient-specific predictive VAR models, our model demonstrated higher accuracy in predicting future observations in terms of both point and interval estimates due to the pooling effect of the hierarchical model specification. In addition, by adopting an elastic-net prior, our model offers greater interpretability about the associations between variables of interest on both the population level and the patient level, as well as between-patient heterogeneity. We apply the model to two examples: 1) predicting substance use craving, negative affect and tobacco use among college students, and 2) predicting functional somatic symptoms and psychological discomforts.

Original languageEnglish (US)
Article numbere0208082
JournalPloS one
Volume13
Issue number12
DOIs
StatePublished - Dec 1 2018

Fingerprint

Databases
Psychology
Delivery of Health Care
health services
Health
Tobacco Use
craving
tobacco use
Tobacco
college students
Time series
Students
signs and symptoms (animals and humans)
time series analysis
Specifications
Population
history
Medically Unexplained Symptoms
Craving

All Science Journal Classification (ASJC) codes

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)

Cite this

Lu, Feihan ; Zheng, Yao ; Cleveland, III, Hobart H. ; Burton, Chris ; Madigan, David. / Bayesian hierarchical vector autoregressive models for patient-level predictive modeling. In: PloS one. 2018 ; Vol. 13, No. 12.
@article{801ff8180500427daa64310debe1fb4a,
title = "Bayesian hierarchical vector autoregressive models for patient-level predictive modeling",
abstract = "Predicting health outcomes from longitudinal health histories is of central importance to healthcare. Observational healthcare databases such as patient diary databases provide a rich resource for patient-level predictive modeling. In this paper, we propose a Bayesian hierarchical vector autoregressive (VAR) model to predict medical and psychological conditions using multivariate time series data. Compared to the existing patient-specific predictive VAR models, our model demonstrated higher accuracy in predicting future observations in terms of both point and interval estimates due to the pooling effect of the hierarchical model specification. In addition, by adopting an elastic-net prior, our model offers greater interpretability about the associations between variables of interest on both the population level and the patient level, as well as between-patient heterogeneity. We apply the model to two examples: 1) predicting substance use craving, negative affect and tobacco use among college students, and 2) predicting functional somatic symptoms and psychological discomforts.",
author = "Feihan Lu and Yao Zheng and {Cleveland, III}, {Hobart H.} and Chris Burton and David Madigan",
year = "2018",
month = "12",
day = "1",
doi = "10.1371/journal.pone.0208082",
language = "English (US)",
volume = "13",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "12",

}

Bayesian hierarchical vector autoregressive models for patient-level predictive modeling. / Lu, Feihan; Zheng, Yao; Cleveland, III, Hobart H.; Burton, Chris; Madigan, David.

In: PloS one, Vol. 13, No. 12, e0208082, 01.12.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Bayesian hierarchical vector autoregressive models for patient-level predictive modeling

AU - Lu, Feihan

AU - Zheng, Yao

AU - Cleveland, III, Hobart H.

AU - Burton, Chris

AU - Madigan, David

PY - 2018/12/1

Y1 - 2018/12/1

N2 - Predicting health outcomes from longitudinal health histories is of central importance to healthcare. Observational healthcare databases such as patient diary databases provide a rich resource for patient-level predictive modeling. In this paper, we propose a Bayesian hierarchical vector autoregressive (VAR) model to predict medical and psychological conditions using multivariate time series data. Compared to the existing patient-specific predictive VAR models, our model demonstrated higher accuracy in predicting future observations in terms of both point and interval estimates due to the pooling effect of the hierarchical model specification. In addition, by adopting an elastic-net prior, our model offers greater interpretability about the associations between variables of interest on both the population level and the patient level, as well as between-patient heterogeneity. We apply the model to two examples: 1) predicting substance use craving, negative affect and tobacco use among college students, and 2) predicting functional somatic symptoms and psychological discomforts.

AB - Predicting health outcomes from longitudinal health histories is of central importance to healthcare. Observational healthcare databases such as patient diary databases provide a rich resource for patient-level predictive modeling. In this paper, we propose a Bayesian hierarchical vector autoregressive (VAR) model to predict medical and psychological conditions using multivariate time series data. Compared to the existing patient-specific predictive VAR models, our model demonstrated higher accuracy in predicting future observations in terms of both point and interval estimates due to the pooling effect of the hierarchical model specification. In addition, by adopting an elastic-net prior, our model offers greater interpretability about the associations between variables of interest on both the population level and the patient level, as well as between-patient heterogeneity. We apply the model to two examples: 1) predicting substance use craving, negative affect and tobacco use among college students, and 2) predicting functional somatic symptoms and psychological discomforts.

UR - http://www.scopus.com/inward/record.url?scp=85058440115&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058440115&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0208082

DO - 10.1371/journal.pone.0208082

M3 - Article

VL - 13

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 12

M1 - e0208082

ER -