This project is a real-world application demonstrating how data science and ML can be used to optimize human performance, rather than replace it, in the context of organ donation.
The work reframes the traditional objective of increasing authorization rates into a more actionable question: how to better align staff strengths with specific clinical and family contexts. Drawing on the economic concept of match quality, a Match Quality (MQ) Score was developed using historical performance data across patient, family, hospital, and coordinator attributes. This score informed the evolution of a deployment tool that supports assigning coordinators to cases where they are most likely to succeed.
Pre- and post-implementation analyses demonstrated substantial improvements in authorization rates across multiple dimensions, including patient demographics, hospital familiarity, and case characteristics. Machine learning models were later introduced to refine feature weighting, identify strong and weak signals, and improve predictive accuracy, with transparency around model limitations.
The results show that responsible use of AI—grounded in ethics, domain expertise, and leadership—can unlock existing human potential, improve equity, and deliver measurable impact. Importantly, the approach positions AI as a decision-support mechanism, reinforcing rather than substituting strong leadership and professional judgment.