AI Uncovers Racial Disparities in Organ Transplant Allocation

Background on Organ Transplantation

The field of organ transplantation has come a long way since the world's first successful transplant in 1954 at Brigham and Women's Hospital. Over the past seven decades, advancements in immunosuppressing drugs and organ matching systems have led to over 1 million organ transplants being performed in the United States alone 1

1

2

2

. However, despite these achievements, the demand for organs still far outweighs the supply, with over 100,000 people currently on the national transplant waiting list and 13 people dying each day while waiting for a transplant.

AI-Powered Analysis Reveals Disparities

Researchers from MIT and Massachusetts General Hospital have recently uncovered concerning disparities in the organ allocation process using advanced computational models. Their study, presented at the ACM Conference on Fairness, Accountability, and Transparency (FAccT) in Athens, Greece, focused on the final stage of organ allocation: when physicians decide whether to accept or reject an organ offer on behalf of a patient 1

1

2

2

.

The team analyzed transplantation data from over 160,000 transplant candidates in the Scientific Registry of Transplant Recipients (SRTR) between 2010 and 2020. Their findings revealed that physicians were overall less likely to accept liver and lung offers for Black candidates, creating additional barriers in the organ allocation process 1

1

2

2

.

Key Findings

1. For liver transplants, Black patients had 7% lower odds of offer acceptance compared to white patients 1

   1

   2

   2

   .
2. The disparity was even more pronounced for lung transplants, with Black patients having 20% lower odds of offer acceptance than white patients with similar characteristics 1

   1

   2

   2

   .
3. Offers were more likely to be accepted if the donor and candidate were of the same race, a trend described as "concerning" given historical inequities in organ procurement 1

   1

   2

   2

   .

Implications and Potential Factors

While the data doesn't necessarily point to clinician bias as the main influence, the researchers suggest that there could be clinical conditions more common among Black patients that aren't accounted for in the wait-list system 1

1

2

2

. This oversight could create obstacles in the process even if the system itself is designed to be unbiased.

The study also highlighted the high variability in offer acceptance and risk tolerances among transplant centers as a potential complicating factor. A referenced 2020 paper in JAMA Cardiology concluded that patients listed at centers with lower offer acceptance rates have a higher likelihood of mortality 1

1

2

2

.

Addressing the Gap with AI and Data

To address these disparities, the researchers have been working on innovative solutions. Last year, they released the Organ Retrieval and Collection of Health Information for Donation (ORCHID) dataset, the first multi-center dataset describing the performance of organ procurement organizations (OPOs) 1

1

2

2

. This 10-year dataset aims to facilitate research addressing bias in organ procurement.

The Importance of Long-Term Commitment in Clinical AI Research

Hammaad Adam, co-first author and recent MIT PhD graduate, emphasized the importance of long-term commitment in clinical AI research. Despite the challenges posed by the bureaucratic and interdisciplinary nature of such projects, Adam dedicated his entire PhD to this work 1

1

2

2

.

He advises graduate students interested in clinical AI research to free themselves from the pressure of frequent publishing cycles, stating, "It's OK if these collaborations take a while. I made the conscious choice a few years ago and I was happy doing that work" 1

1

2

2

.

This research demonstrates the potential of AI and data analysis in uncovering and addressing healthcare disparities, paving the way for more equitable organ allocation systems in the future.