Making Large Cox’s Proportional Hazard Models Tractable in Bayesian Networks

Jidapa Kraisangka, Marek J. Druzdzel
Proceedings of the Eighth International Conference on Probabilistic Graphical Models, PMLR 52:252-263, 2016.

Abstract

Cox’s proportional hazard (CPH) model is a statistical technique that captures the interaction between a set of risk factors and an effect variable. While the CPH model is popular in survival analysis, Bayesian networks offer an attractive alternative that is intuitive, general, theoretically sound, and avoids CPH model’s restrictive assumptions. Existing CPH models are a great source of existing knowledge that can be reused in Bayesian networks. The main problem with applying Bayesian networks to survival analysis is their exponential growth in complexity as the number of risk factors increases. It is not uncommon to see complex CPH models with as many as 20 risk factors. Our paper focuses on making large survival analysis models derived from the CPH model tractable in Bayesian networks. We evaluate the effect of two complexity reduction techniques: (1) parent divorcing, and (2) removing less important risk factors based on the accuracy of the resulting models.

Cite this Paper

BibTeX
@InProceedings{pmlr-v52-kraisangka16, title = {Making Large Cox's Proportional Hazard Models Tractable in {B}ayesian Networks}, author = {Kraisangka, Jidapa and Druzdzel, Marek J.}, booktitle = {Proceedings of the Eighth International Conference on Probabilistic Graphical Models}, pages = {252--263}, year = {2016}, editor = {Antonucci, Alessandro and Corani, Giorgio and Campos}, Cassio Polpo}, volume = {52}, series = {Proceedings of Machine Learning Research}, address = {Lugano, Switzerland}, month = {06--09 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v52/kraisangka16.pdf}, url = {https://proceedings.mlr.press/v52/kraisangka16.html}, abstract = {Cox’s proportional hazard (CPH) model is a statistical technique that captures the interaction between a set of risk factors and an effect variable. While the CPH model is popular in survival analysis, Bayesian networks offer an attractive alternative that is intuitive, general, theoretically sound, and avoids CPH model’s restrictive assumptions. Existing CPH models are a great source of existing knowledge that can be reused in Bayesian networks. The main problem with applying Bayesian networks to survival analysis is their exponential growth in complexity as the number of risk factors increases. It is not uncommon to see complex CPH models with as many as 20 risk factors. Our paper focuses on making large survival analysis models derived from the CPH model tractable in Bayesian networks. We evaluate the effect of two complexity reduction techniques: (1) parent divorcing, and (2) removing less important risk factors based on the accuracy of the resulting models.} }
Endnote
%0 Conference Paper %T Making Large Cox’s Proportional Hazard Models Tractable in Bayesian Networks %A Jidapa Kraisangka %A Marek J. Druzdzel %B Proceedings of the Eighth International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2016 %E Alessandro Antonucci %E Giorgio Corani %E Cassio Polpo Campos} %F pmlr-v52-kraisangka16 %I PMLR %P 252--263 %U https://proceedings.mlr.press/v52/kraisangka16.html %V 52 %X Cox’s proportional hazard (CPH) model is a statistical technique that captures the interaction between a set of risk factors and an effect variable. While the CPH model is popular in survival analysis, Bayesian networks offer an attractive alternative that is intuitive, general, theoretically sound, and avoids CPH model’s restrictive assumptions. Existing CPH models are a great source of existing knowledge that can be reused in Bayesian networks. The main problem with applying Bayesian networks to survival analysis is their exponential growth in complexity as the number of risk factors increases. It is not uncommon to see complex CPH models with as many as 20 risk factors. Our paper focuses on making large survival analysis models derived from the CPH model tractable in Bayesian networks. We evaluate the effect of two complexity reduction techniques: (1) parent divorcing, and (2) removing less important risk factors based on the accuracy of the resulting models.
RIS
TY - CPAPER TI - Making Large Cox’s Proportional Hazard Models Tractable in Bayesian Networks AU - Jidapa Kraisangka AU - Marek J. Druzdzel BT - Proceedings of the Eighth International Conference on Probabilistic Graphical Models DA - 2016/08/15 ED - Alessandro Antonucci ED - Giorgio Corani ED - Cassio Polpo Campos} ID - pmlr-v52-kraisangka16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 52 SP - 252 EP - 263 L1 - http://proceedings.mlr.press/v52/kraisangka16.pdf UR - https://proceedings.mlr.press/v52/kraisangka16.html AB - Cox’s proportional hazard (CPH) model is a statistical technique that captures the interaction between a set of risk factors and an effect variable. While the CPH model is popular in survival analysis, Bayesian networks offer an attractive alternative that is intuitive, general, theoretically sound, and avoids CPH model’s restrictive assumptions. Existing CPH models are a great source of existing knowledge that can be reused in Bayesian networks. The main problem with applying Bayesian networks to survival analysis is their exponential growth in complexity as the number of risk factors increases. It is not uncommon to see complex CPH models with as many as 20 risk factors. Our paper focuses on making large survival analysis models derived from the CPH model tractable in Bayesian networks. We evaluate the effect of two complexity reduction techniques: (1) parent divorcing, and (2) removing less important risk factors based on the accuracy of the resulting models. ER -
APA
Kraisangka, J. & Druzdzel, M.J.. (2016). Making Large Cox’s Proportional Hazard Models Tractable in Bayesian Networks. Proceedings of the Eighth International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 52:252-263 Available from https://proceedings.mlr.press/v52/kraisangka16.html.

Related Material