Skip to main content

How AI Changed Organ Donation in the US

Article Link:

This article talks about how Artificial Intelligence is using matching markets to revolutionize the kidney donation process. Previously, people would depend on finding willing living donors that they knew personally that were either friends or family members. But finding a willing kidney donor is the easy part since humans are born with 2 kidneys and only need 1 to be functional. What’s difficult is actually making sure that the donor is able to donate the kidney to the recipient. The first hospital to begin the practice of paired kidney transplant was in South Korea in the 1990s. They matched two separate donors and patients together. After that, doctors used to spend grueling amounts of time going through biological data to manually match patients with their list of donors. In the early 2000s, Alan and Michael Rees helped to develop AI to analyze the data given so that their results would be more accurate than what humans could do. Now, the biggest issue in the organ donor and recipient matching is trying to understand the moral dilemma of deciding which person deserves to get a donation over another. Researchers are trying to teach machines to understand these sorts of moral issues like a human would. 


This image provided by the article is a bipartite graph that represents a perfect matching. As we can see, Donor A matches with Patient B and Donor B matches with Patient A. Matches are made primarily on biological suitability (blood type, health, tissue matching, etc.), time the patient has been on the waiting list, age, and whether the person who needs a kidney has been a living organ donor in the past. Hardest to match patients get the priority. However, it is less likely that a small network like this would have a perfect matching like this because so much goes into the matching criteria. A thicker the network, which has more people, would be more likely to be a perfect matching. However, the organ donor and recipient network will probably never be a perfect matching because of the many requirements that come with being a suitable match and the moral dilemma that arises of choosing which person deserves a donation more than another. This is why researchers are putting in the effort to have AI understand these moral dilemmas instead of using pure computation alone to make these decisions. In real life, matching isn’t that simple. There isn’t always one right way to create these matchings. But, with AI, we are trying to figure out a way to reach a perfect matching.


Leave a Reply

Blogging Calendar

September 2018