Management of a Chronically Ill Population: An Operations Approach to Liver Cancer Screening.

Screening;Disease Modeling;Reinforcement Learning;Restless Bandits;Engineering (General);Engineering;Industrial and Operations Engineering

Lee, ElliotShi, Cong ;

University of Michigan

关键词: Screening;    Disease Modeling;    Reinforcement Learning;    Restless Bandits;    Engineering (General);    Engineering;    Industrial and Operations Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/133270/elliotdl_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

We study how to perform medical surveillance of a population living with a chronic disease from an operations perspective. Our approach to the screening problem is the first to combine aspects of patient specific risk factors, heterogenous disease progression, as well limited screening resources shared by the population. Using clinical data from liver cancer as a motivating example, we (1) provide a new characterization of individualized risk for liver cancer through a nested case-control match study, then (2) demonstrate the utility of that individual biological information in screening decisions through the design and testing of reinforcement learning techniques, and then (3) model the problem as a family of restless bandits to gain structural insights into the problem, as well as derive an optimal policy to screen patients. Ultimately, we provide novel methods of screening a chronically ill population which are superior to current practice by adopting principles from a broad spectrum of operations methods.