Machine Learning approach to Predicting Stem-Cell Donor Availability
0. AbstractThe success of Unrelated Donor stem-cell transplants depends not only on finding genetically matched donors but also on donor availability. On average 50% of potential donors in the NMDP database are unavailable for a variety of reasons, after initially matching a patient, with significant variations in availability among subgroups (e.g., by race or age). Several studies have established univariate donor characteristics associated with availability. Individual consideration of each applicable characteristic is laborious. Extrapolating group averages to individual donor level tends to be highly inaccurate. In the current environment with enhanced donor data collection, we can make better estimates of individual donor availability. In this study, we propose a Machine Learning based approach to predict availability of every registered donor, to be used during donor selection and reduce the time taken to complete a transplant.