A Holistic Approach for the Simulation, Evaluation, and Design of Organ Transplantation Policies in the U.S.

ABSTRACT

In 2025, nearly 75,000 patients in the U.S. were added to the national organ transplant waitlist, while only 49,000 transplants were performed. When an organ is donated, there are many patients across the country who could receive it. Organ allocation policies, set by the Organ Procurement and Transplantation Network (OPTN), determine how these patients are prioritized. OPTN policymakers seek to make the best use of a limited supply of transplantable organs—maximizing added years of life, utilizing as many organs as possible, and distributing benefits fairly. However, designing these policies is challenging, and the analytical tools that support the policymakers have advanced little in decades. This thesis develops and demonstrates new computational methods to address these challenges.

First, an improved algorithm is developed for performing simulation of the transplant system, which is critical to forecasting the effects of policy changes. With the new algorithm, simulation runtime is reduced from nearly 7 hr to approximately 15 s for an existing kidney/pancreas simulation model. Whereas traditional policy development was limited to analyzing a handful of policies over several months, this advancement enables analysis of tens of thousands of policies in days. An interactive website is created to enable policymakers to explore results and design policies. These tools are applied to characterize and optimize over a space of kidney allocation policies under consideration by the OPTN.

Broad policy simulation is confined to a parameterized space of policies. To study fundamental limits in organ allocation, optimal assignment problems are used to analyze underlying feasible donor-recipient assignments. This is applied to longstanding concerns in kidney policy regarding logistical efficiency and geographic disparities in transplantation access. After developing a national-scale model of kidney transportation logistics, the analysis shows regional kidney transplant counts—ranging from 40% below to 60% above expected levels in 2024—could be brought within 3.5% of expected levels, while reducing transportation time by 45%, air cargo usage by 15%, and the number of shipments longer than 10 hr by 70%.

Insights from optimal assignment analyses can be used to improve the space of policies considered during broad simulation by guiding the computational design of scoring formulas. A gradient-free optimization approach over flexible, nonparametric scoring functions is used to design formulas that achieve policy objectives. Applied to lung transplantation, this approach identifies adjustments to a recently approved policy that reduces the range in transplant rate across patient blood type and chest cavity size by 39% in simulation.

Finally, a new simulator of the U.S. transplant system is developed. Whereas existing simulators are limited to specific organ types, the new simulator includes all major organ types, unifying simulation of the system. The high-speed simulation algorithm is extended to the multi-organ setting, enabling execution in minutes. The new simulator is validated against historical data from the OPTN.

Together, these contributions form a holistic approach to the design of organ allocation policy. These techniques can help policymakers translate their goals into concrete, effective, evidence-based decisions in a setting with serious consequences for tens of thousands of patients every year.

BIOGRAPHY

Elijah Pivo is a doctoral candidate in the Institute for Data, Systems, and Society at the Massachusetts Institute of Technology, where he is advised by Prof. Dimitris Bertsimas. His research focuses on computational modeling and optimization for the U.S. organ transplant system. He received his B.S. in Electrical Engineering and Computer Engineering with a minor in robotics from Johns Hopkins University in 2018, where he was awarded the William H. Huggins Award in Computer Engineering. He is a recipient of the National Science Foundation Graduate Research Fellowship, the Michael Hammer Fellowship, and the MIT HEALS Graduate Fellowship.

COMMITTEE

Dimitris Bertsimas, David Gamarnik, Nikolaos Trichakis

EVENT INFORMATION

Hybrid event. To attend virtually, please contact the IDSS Academic Office (idss_academic_office@mit.edu) for connection information.