Developing New Models to Inform a Proposed Policy for Fair Kidney Transplant Allocation

The U.S. deceased-donor organ allocation system faces significant inefficiencies, including inequitable access, disparities among patients, low organ utilization, and high patient mortality. The Organ Procurement and Transplantation Network (OPTN) is responsible for administering this system, which currently affects over 115,000 patients on the national waitlist for transplants and nearly 1,000,000 patients with chronic diseases that could be effectively treated through transplantation.  Despite efforts to improve allocation outcomes, inefficiencies persist.

Profs. Bertsimas and Trichakis have done prior research on organ allocation. Now, again collaborating with the OPTN, they are turning their focus from lungs to kidneys. They intend to develop new models that will underpin a proposed kidney allocation policy that will improve transplant activities and patient welfare. Specifically, the policy will aim to improve utilization, reduce kidney discard, reduce mortality in the waitlist, and reduce disparities in transplant rates across patient groups.

The research problem poses challenges in modeling patient behavior and choice, as kidney transplant candidates have an alternative treatment available (dialysis), making them selective in accepting or declining organ offers.

The research strategy includes developing three models and an online decision-support dashboard.
- Patient choice model: A model that predicts the probability of a particular patient accepting or declining a particular organ offer.
- Simulation model: A detailed simulation model that makes counter-factual predictions about important outcomes of interest.
- Optimization model: A framework that uses machine learning and optimization models to approximate allocation outcomes.
- Online dashboard: A flexible decision-support tool that enables policymakers to quickly iterate on different policy scenarios and refine their value judgments on relevant tradeoffs.