MIE Seminar: An Application of Markov Decision Process in Healthcare: Mitigating Inequities in Organ Allocation via Revised Health Reporting Frequencies
~~ The Department of Mechanical and Industrial Engineering Presents
Dr. Zeynep Gozde Icten
Topic: An Application of Markov Decision Processes in Healthcare: Mitigating Inequities in Organ Allocation via Revised Health Reporting Frequencies
Date: Friday, March 28, 2014
Time: 10:00am to 11:00am (Refreshments from 9:30am to 10:00am)
Location: 406 Egan Center
Abstract: Due to the scarcity of donated livers, it is critical that the United Network for Organ Sharing (UNOS), the organization responsible for nationwide allocation of donated organs in the US, manage organs in an efficient, effective and equitable way. UNOS prioritizes patients awaiting liver transplantation based primarily on their medical urgency, as measured by their model for end-stage liver disease (MELD) score, and requires each patient to report their MELD score at a frequency that depends on their last reported MELD score (the sicker, the more frequent). As a result of this flexibility, patients may conceal changes in their MELD score and “game” the system. Mitigating the resulting inequity by requiring very frequent updates, however, is impractical and would add to the already significant data processing burden. Using a Markov decision process model parameterized by clinical data and cost-effectiveness analysis, we examine (i) the degree to which an individual patient can benefit from the updating flexibility, and (ii) how the resulting inequities may be mitigated by revising the updating frequencies without significantly adding to the data processing burden. We provide a menu of updating policies that balance inequity and data processing and suggest that requiring the sicker (healthier) patients to update more (less) frequently than they must under the current policy can improve both metrics.This is joint work with Dr Lisa Maillart, Dr Andrew Schaefer, Dr Mark Roberts and Dr Atul Bhandari.
Bio: Zeynep G Icten, PhD holds a Bachelor’s Degree in Industrial Engineering from Bogazici University and a Master’s Degree in Industrial Engineering from University of Pittsburgh. She completed her Doctoral Degree in Industrial Engineering at University of Pittsburgh where her main research interest was decision making under uncertainty with healthcare and maintenance optimization applications. She served as a teaching assistant to a variety of classes and resumed full responsibility to teach a core junior/senior undergraduate class (Probabilistic Methods in Operations Research) for which she was awared an "Outstanding Teaching Award". Following her graduation Dr Icten took a Research Associate of Health Economics position in Modeling and Simulation department of Evidera where her primary research focused on modeling and simulation of pharmacoeconomics problems to generate cost-effectiveness results of various therapeutic interventions and medical devices. She worked with Markov Cohort models, discrete event simulation models, individual patient simulation models and survival models. She has expertise in multiple disease areas including multiple myeloma, rheumatoid arthritis, cardiovascular diseases, overactive bladder disease and Gaucher disease. Currently, Dr Icten hols a Research Associate position at GNS Healthcare where her research focuses on Bayesian predictive modeling, machine learning and big data analysis. She has published in IIE Transactions, Journal of Risk and Reliability, has multiple proceedings, has a book chapter in Wiley Encyclopedia of Operations Research and Management Science, and refereed for Journal of Risk and Reliability. She has presented in INFORMS, IERC, SMDM and ISPOR.
Friday, March 28, 2014 at 10:00am to 11:00am
Egan Research Center, 406
120 Forsyth Street, Boston, MA, Boston