Understanding disease processes by partitioned dynamic Bayesian networks

Marcos L. P. Bueno*, Arjen Hommersom, Peter J. F. Lucas, Martijn Lappenschaar, Joost G. E. Janzing

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

For many clinical problems in patients the underlying pathophysiological process changes in the course of time as a result of medical interventions. In model building for such problems, the typical scarcity of data in a clinical setting has been often compensated by utilizing time homogeneous models, such as dynamic Bayesian networks. As a consequence, the specificities of the underlying process are lost in the obtained models. In the current work, we propose the new concept of partitioned dynamic Bayesian networks to capture distribution regime changes, i.e. time non-homogeneity, benefiting from an intuitive and compact representation with the solid theoretical foundation of Bayesian network models. In order to balance specificity and simplicity in real-world scenarios, we propose a heuristic algorithm to search and learn these non-homogeneous models taking into account a preference for less complex models. An extensive set of experiments were ran, in which simulating experiments show that the heuristic algorithm was capable of constructing well-suited solutions, in terms of goodness of fit and statistical distance to the original distributions, in consonance with the underlying processes that generated data, whether it was homogeneous or non-homogeneous. Finally, a study case on psychotic depression was conducted using non-homogeneous models learned by the heuristic, leading to insightful answers for clinically relevant questions concerning the dynamics of this mental disorder. (C) 2016 Elsevier Inc. All rights reserved.

Original languageEnglish
Pages (from-to)283-297
Number of pages15
JournalJournal of Biomedical Informatics
Volume61
DOIs
Publication statusPublished - Jun 2016

Keywords

  • Probabilistic graphical models
  • Dynamic Bayesian networks
  • Non-homogeneous stochastic processes
  • Heuristic algorithm
  • Multivariate time series
  • Psychotic depression
  • RATING-SCALE
  • TIME-SERIES
  • DEPRESSION
  • DIAGNOSIS
  • MODEL
  • INFORMATION

Cite this

Bueno, Marcos L. P. ; Hommersom, Arjen ; Lucas, Peter J. F. ; Lappenschaar, Martijn ; Janzing, Joost G. E. / Understanding disease processes by partitioned dynamic Bayesian networks. In: Journal of Biomedical Informatics. 2016 ; Vol. 61. pp. 283-297.
@article{817ed4a3a2f04ca5ac749894ca386d9c,
title = "Understanding disease processes by partitioned dynamic Bayesian networks",
abstract = "For many clinical problems in patients the underlying pathophysiological process changes in the course of time as a result of medical interventions. In model building for such problems, the typical scarcity of data in a clinical setting has been often compensated by utilizing time homogeneous models, such as dynamic Bayesian networks. As a consequence, the specificities of the underlying process are lost in the obtained models. In the current work, we propose the new concept of partitioned dynamic Bayesian networks to capture distribution regime changes, i.e. time non-homogeneity, benefiting from an intuitive and compact representation with the solid theoretical foundation of Bayesian network models. In order to balance specificity and simplicity in real-world scenarios, we propose a heuristic algorithm to search and learn these non-homogeneous models taking into account a preference for less complex models. An extensive set of experiments were ran, in which simulating experiments show that the heuristic algorithm was capable of constructing well-suited solutions, in terms of goodness of fit and statistical distance to the original distributions, in consonance with the underlying processes that generated data, whether it was homogeneous or non-homogeneous. Finally, a study case on psychotic depression was conducted using non-homogeneous models learned by the heuristic, leading to insightful answers for clinically relevant questions concerning the dynamics of this mental disorder. (C) 2016 Elsevier Inc. All rights reserved.",
keywords = "Probabilistic graphical models, Dynamic Bayesian networks, Non-homogeneous stochastic processes, Heuristic algorithm, Multivariate time series, Psychotic depression, RATING-SCALE, TIME-SERIES, DEPRESSION, DIAGNOSIS, MODEL, INFORMATION",
author = "Bueno, {Marcos L. P.} and Arjen Hommersom and Lucas, {Peter J. F.} and Martijn Lappenschaar and Janzing, {Joost G. E.}",
year = "2016",
month = "6",
doi = "10.1016/j.jbi.2016.05.003",
language = "English",
volume = "61",
pages = "283--297",
journal = "Journal of Biomedical Informatics",
issn = "1532-0464",
publisher = "ACADEMIC PRESS INC ELSEVIER SCIENCE",

}

Understanding disease processes by partitioned dynamic Bayesian networks. / Bueno, Marcos L. P.; Hommersom, Arjen; Lucas, Peter J. F.; Lappenschaar, Martijn; Janzing, Joost G. E.

In: Journal of Biomedical Informatics, Vol. 61, 06.2016, p. 283-297.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Understanding disease processes by partitioned dynamic Bayesian networks

AU - Bueno, Marcos L. P.

AU - Hommersom, Arjen

AU - Lucas, Peter J. F.

AU - Lappenschaar, Martijn

AU - Janzing, Joost G. E.

PY - 2016/6

Y1 - 2016/6

N2 - For many clinical problems in patients the underlying pathophysiological process changes in the course of time as a result of medical interventions. In model building for such problems, the typical scarcity of data in a clinical setting has been often compensated by utilizing time homogeneous models, such as dynamic Bayesian networks. As a consequence, the specificities of the underlying process are lost in the obtained models. In the current work, we propose the new concept of partitioned dynamic Bayesian networks to capture distribution regime changes, i.e. time non-homogeneity, benefiting from an intuitive and compact representation with the solid theoretical foundation of Bayesian network models. In order to balance specificity and simplicity in real-world scenarios, we propose a heuristic algorithm to search and learn these non-homogeneous models taking into account a preference for less complex models. An extensive set of experiments were ran, in which simulating experiments show that the heuristic algorithm was capable of constructing well-suited solutions, in terms of goodness of fit and statistical distance to the original distributions, in consonance with the underlying processes that generated data, whether it was homogeneous or non-homogeneous. Finally, a study case on psychotic depression was conducted using non-homogeneous models learned by the heuristic, leading to insightful answers for clinically relevant questions concerning the dynamics of this mental disorder. (C) 2016 Elsevier Inc. All rights reserved.

AB - For many clinical problems in patients the underlying pathophysiological process changes in the course of time as a result of medical interventions. In model building for such problems, the typical scarcity of data in a clinical setting has been often compensated by utilizing time homogeneous models, such as dynamic Bayesian networks. As a consequence, the specificities of the underlying process are lost in the obtained models. In the current work, we propose the new concept of partitioned dynamic Bayesian networks to capture distribution regime changes, i.e. time non-homogeneity, benefiting from an intuitive and compact representation with the solid theoretical foundation of Bayesian network models. In order to balance specificity and simplicity in real-world scenarios, we propose a heuristic algorithm to search and learn these non-homogeneous models taking into account a preference for less complex models. An extensive set of experiments were ran, in which simulating experiments show that the heuristic algorithm was capable of constructing well-suited solutions, in terms of goodness of fit and statistical distance to the original distributions, in consonance with the underlying processes that generated data, whether it was homogeneous or non-homogeneous. Finally, a study case on psychotic depression was conducted using non-homogeneous models learned by the heuristic, leading to insightful answers for clinically relevant questions concerning the dynamics of this mental disorder. (C) 2016 Elsevier Inc. All rights reserved.

KW - Probabilistic graphical models

KW - Dynamic Bayesian networks

KW - Non-homogeneous stochastic processes

KW - Heuristic algorithm

KW - Multivariate time series

KW - Psychotic depression

KW - RATING-SCALE

KW - TIME-SERIES

KW - DEPRESSION

KW - DIAGNOSIS

KW - MODEL

KW - INFORMATION

U2 - 10.1016/j.jbi.2016.05.003

DO - 10.1016/j.jbi.2016.05.003

M3 - Article

VL - 61

SP - 283

EP - 297

JO - Journal of Biomedical Informatics

JF - Journal of Biomedical Informatics

SN - 1532-0464

ER -