Oral Defense of Doctoral Dissertation – Computational Science and Informatics- Modeling, Simulation, and Analysis of the US Organ Transplant System – Christine Harvey
Notice and Invitation
Oral Defense of Doctoral Dissertation
Doctor of Philosophy in Computational Sciences and Informatics
Department of Computational and Data Sciences
College of Science
George Mason University
Bachelor of Science, Stockton University, 2011
Master of Science, Stockton University, 2013
Modeling, Simulation, and Analysis of the US Organ Transplant System
Tuesday, October 29, 2:00 p.m.
Exploratory Hall, Room 3301
All are invited to attend.
Robert Weigel, Dissertation Director
Andrew Crooks, Committee Chair
Analysis, modeling, and simulation of organ transplantation and donation can enhance the understanding of this complex system and guide strategic policy improvements. Four major research questions are addressed in this work: (1) how can we further enable data-driven research of the transplant system for future scientists?; (2) what demographic factors influence donations and access to transplantation?; (3) how do laws and policies affect organ donations?; and (4) how do certain patient advantages impact the overall system as well as those lacking advantages?
A data pipeline and associated software were developed and published that address how to further data-driven research of the transplant system for future scientists. This software simplifies access to and analysis of data from proprietary Organ Procurement and Transplantation Network (OPTN) Standard Transplant Analysis and Research (STAR) files to an open-source database format. These files contain data on every organ donor, waitlist registrant, and transplant recipient since 1987 in the US. This data pipeline directly facilitated the next phase of research which involved performing an analysis of the transplant system using this dataset. The exploratory data analysis scales transplant data to the relative populations to gain a better understanding of the differences between demographic groups and reveals important differences across education levels, gender, race, and ethnicity.
Demographic factors influencing organ donation and access to transplants are analyzed in this work through exploratory visualizations and predictive modeling. A visual exploratory analysis is presented which examines demographic features of organ donors and highlights differences in intersectional data across the population of donors compared to the relative population described by the US Census. Additionally, a random forest model is used to determine the features of patients on the waitlist for a kidney transplant to determine if certain attributes may inadvertently drive the allocation system. This model predicts patient outcomes based on features represented in the model with an accuracy above the zero-rule baseline. The analysis found that patient age, year of listing, body weight, and zip code are important factors in determining a patient’s outcome – other demographic factors such as race and gender were not important prediction features.
State and local laws, policies, and their impact on organ donation are evaluated through a statistical analysis that compares donations after the implementation of a policy to areas without the policy implementation. A database of state and local laws and policies and the years of implementation was developed to compare donations across the country. The results demonstrated that some policies can be correlated with a change in donation, but only for certain demographic subgroups in a population.
Finally, I built discrete event simulation models of a representative patient population to determine the impact of changes to the transplant system that can not be easily demonstrated in the real world. A transplant process model was developed to determine how increasing living and deceased donation overall and within racial sub-groups would impact the number of donors each year. Additionally, an agent-based queuing model was used to understand the impact of allowing patients to register within more than one area. This model provides a valuable tool for examining policy changes that shows the global and local impacts of multiple listing. The analysis found that multiply listed patients have improved access to transplants and are less likely to die while waiting for a transplant.