A dynamic Bayesian network to represent discrete duration models

Originally devoted to specific applications such as biology, medicine and demography, duration models are now widely used in economy, finance or reliability. Recent works in various fields of application have shown the relevancy of using Bayesian networks to model complex systems, namely stochastic systems with an underlying distribution that does not fit to a well-known parametric form. In this paper, the description of a specific dynamic Bayesian network, referred to as Graphical Duration Model (GDM), is given. A GDM aims to represent a wide range of duration models. Its structure allows especially to fit multi-state systems featuring complex sojourn-time distributions and contextual dependencies. To that end, a duration variable is explicitly introduced to the state transition model which is classically represented by a Markov chain. A recursive algorithm efficiently perform inference in this model is derived along with its proof of correctness and space and time complexity studies. Finally, this approach is illustrated with an application in survival analysis in which the proposed model is compared with the commonly used Markov chain modelling.

[1]  F. J. Díez,et al.  Operating with potentials of discrete variables , 2007, Int. J. Approx. Reason..

[2]  Stuart J. Russell,et al.  Dynamic bayesian networks: representation, inference and learning , 2002 .

[3]  Michael I. Jordan.  Learning in Graphical Models , 1999, NATO ASI Series.

[4]  J. Kalbfleisch,et al.  The Statistical Analysis of Failure Time Data , 1980 .

[5]  Xavier Boyen,et al.  Tractable Inference for Complex Stochastic Processes , 1998, UAI.

[6]  D. Cox.  Regression Models and Life-Tables , 1972 .

[7]  Philippe Weber,et al.  Reliability modelling with dynamic bayesian networks , 2003 .

[8]  David J. Spiegelhalter,et al.  Local computations with probabilities on graphical structures and their application to expert systems , 1990 .

[9]  F. Cozman,et al.  Generalizing variable elimination in Bayesian networks , 2000 .

[10]  Rina Dechter,et al.  Bucket Elimination: A Unifying Framework for Reasoning , 1999, Artif. Intell..

[11]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems - networks of plausible inference , 1991, Morgan Kaufmann series in representation and reasoning.

[12]  Joanne Bechta Dugan,et al.  A discrete-time Bayesian network reliability modeling and analysis framework , 2005, Reliab. Eng. Syst. Saf..

[13]  J. Kalbfleisch,et al.  The Statistical Analysis of Failure Time Data: Kalbfleisch/The Statistical , 2002 .

[14]  Olivier Gaudoin,et al.  A SURVEY ON DISCRETE LIFETIME DISTRIBUTIONS , 2003 .

[15]  Finn Verner Jensen,et al.  Introduction to Bayesian Networks , 2008, Innovations in Bayesian Networks.