Title: Preference Summaries for Stochastic Tree Rollback
Authors: Gordon B. Hazen, Northwestern University and Jayavel Sounderpandian, University of Wisconsin-Parkside
Date: June1998
Status: working paper

A stochastic tree is a convenient structure to represent the future health process of a patient, and it can be used to make complex medical decisions or to carry out cost effectiveness analyses of expensive medical treatments. Several types of von Neumann utility functions defined on stochastic trees are recursive in that they allow rollback of the stochastic tree, as in the case of decision trees. Most recursive utility functions also admit preference summaries that can be used to decompose a stochastic tree into preference elements and probabilistic elements. Through an example, we describe this decomposition and subsequent utility computations.

Click here to access a pdf version of this paper.

Back to the Decision Analysis Working Paper Index