Markov versus Medical Markov Modeling – Contrasts and Refinements
Monday, February 13th | 11:30 am-12:20 pm | Emerging Technologies Building Room:1020
Abstract: Amounting to several hundred publications per year in the medical literature, Markov models are widely employed in cost-effectiveness analysis of healthcare interventions. Nearly all medical Markov models consist of transient chains solved by simulation or cohort analysis. This is in contrast to ergotic models whose steady-state solution methods are typically taught to operations research graduate students. In this talk, I focus on two modeling strategies that might arise in a medical context – the decomposition of Markov chains into simper components, and the explicit treatment of population features of such models.
Regarding decomposition, it can occur that a Markov model does not assume any probabilistic dependence among its components. In this case, a standard cohort analysis may be decomposed into several independent cohort analyses, one for each component, and the results may be combined to produce desired expected costs and quality-adjusted life years (QALYs). Graphical depiction of the simple components that comprise a model reduces model complexity, makes model formulation easier and more transparent, and thereby facilitates peer inspection and critique.
We also examine population-level Markov models of non-interacting individuals using techniques from the literature on stochastic networks. A key construct from this literature is the notion of population equilibrium. We also examine potentially relevant measures of health-related outcome, exploring how they differ from each other and from individual-level measures such as QALYs. As we illustrate, population modeling can provide a more refined picture of outcome. For instance, a beneficial intervention might increase population size but result on average in less healthy individuals.
Regarding decomposition, it can occur that a Markov model does not assume any probabilistic dependence among its components. In this case, a standard cohort analysis may be decomposed into several independent cohort analyses, one for each component, and the results may be combined to produce desired expected costs and quality-adjusted life years (QALYs). Graphical depiction of the simple components that comprise a model reduces model complexity, makes model formulation easier and more transparent, and thereby facilitates peer inspection and critique.
We also examine population-level Markov models of non-interacting individuals using techniques from the literature on stochastic networks. A key construct from this literature is the notion of population equilibrium. We also examine potentially relevant measures of health-related outcome, exploring how they differ from each other and from individual-level measures such as QALYs. As we illustrate, population modeling can provide a more refined picture of outcome. For instance, a beneficial intervention might increase population size but result on average in less healthy individuals.
Bio: Gordon Hazen is professor of Industrial Engineering and Management Sciences at Northwestern University. His research interests include decision analysis methodology, utility and preference theory, medical decision analysis, and cost-effectiveness analysis of medical treatment decisions. He has published in leading journals including Management Science, Operations Research, and Medical Decision Making. He is the outgoing Area Editor for decision analysis at Operations Research, and is a member of the editorial board of the INFORMS journal Decision Analysis.