| The risk of colorectal cancer (CRC), the second deadliest cancer in the U.S., can be significantly reduced by periodic colonoscopy screening. Although there are published guidelines on the frequency of colonoscopy screening, there exists significant controversy on how often to screen patients, as reflected by the discrepancy between clinical practice and guidelines. Therefore, this dissertation focuses on modeling the progression of CRC and finding the optimal CRC screening and surveillance policies.;In the first part of this dissertation, we introduce a finite-horizon partially observable Markov decision process (POMDP) optimizing colonoscopy screening policies for CRC prevention and surveillance. In this context, we determine the screening policies that maximize total expected quality-adjusted life years of a particular patient with respect to age, gender and personal CRC history. Our POMDP model also incorporates personal risk of having CRC and adenomatous polyps into the screening decisions, therefore, provides a novel framework towards personalized screening policies. We use clinical data for our numerical experiments and find that more frequent colonoscopy screening than the guidelines is required to achieve maximal benefit from CRC screening. When compared to males, the optimal screening policies recommend females with CRC history undergo colonoscopy more frequently. In contrast, females without CRC history should be screened less frequently than males. This result, which was not recognized before, signifies the role of gender in the optimal CRC screening decisions.;Our POMDP model requires certain inputs including the parameters of metachronous CRC (a second primary CRC) progression. Therefore, in the second part of this dissertation, we describe a simulation-based natural history model for metachronous CRC that estimates these progression parameters. In the third and the fourth parts of this dissertation, we extend the POMDP model in two directions: For the first extension, we consider the costs of CRC screening and treatment while optimizing colonoscopy screening decisions. For the second extension, we adjusted initial POMDP model to consider multiple screening modalities. |
