Document Type


Date of Award


Degree Name

Master of Engineering (MEngr)


<--Please Select Department-->

First Advisor

Dr. Sang Won Yoon


There is ever increasing disparity between number of organs needed for transplantation and numbers available for donation to save lives. As a result, thousands of people die every year waiting for organs. Therefore, it is now more important than ever before to take serious actions to decrease this disparity. One way to bridge gap between organ demand and supply is to increase family consent for organ donation. This research studied the factors associated with family consent. Machine Learning approach had been used in very few literature to understand factors related to family consent. This study uses six Ensemble Machine Learning models to accurately predict family consent outcome (yes/no). All family approaches data between January 2016 and March 2018 from an Organ Procurement Organization (OPO) based in New York city is used to build the family consent prediction model. The experimental results reveals that eXtreme Gradient Boosting (XGB) Machine Learning model performs better than other ensemble models with AUC of 0.8946 and accuracy of 81.7% after normalizing features and using LDA for dimension reduction and then tuning parameters using grid search method. 24 out of 29 features are identied as important features by XGB model. The model is used to calculate probability of consent before approaching family as the values for dierent features are available real-time after patient is referred to OPO for medical evaluation and suitability. The experimental result shows that the accuracy of the model increases from 77.6% to 91.5% as value for factors are added real-time. This model is also used for selecting the best sta for a particular case to approach family based on their past experience. Sta work schedule is incorporated with the model to select the top three sta based on likelihood of getting consent from family for organ donation. This recommendation system can be used as a potential sta dispatch model for OPO to further improve the consent from family for organ donation and save more lives by customizing the sta deployment procedure based on the characteristics of donor referral.

Available for download on Tuesday, April 20, 2021