houston voices

Houston researchers build life-saving decision making mathematical model with big potential

Researchers created a mathematical model that helps transplant centers make decisions about when to move forward with a matching donor and when to wait. This work can potentially help decision making in other industries. Photo via Getty Images

To wait, or not to wait? That is the question — or at least it might be, if you need a kidney transplant.

Nearly 89,000 Americans with chronic kidney disease are on a waitlist for a new organ, and an estimated 13 people die each day while awaiting a transplant. But there are real costs to matching patients with the first donor that becomes available, just as there are equally real costs to having them wait in hopes of finding a better one.

Recently, Rice Business professor Süleyman Kerimov and colleagues at Stanford University and Northwestern University developed a mathematical model that helps clarify when it's best to match patients to donors as quickly as possible and when it's best to wait.

Their findings, which appear in two papers published in Management Science and Operations Research, respectively, could help optimize all manner of matching markets in which participants seek to connect with potential partners based on mutual compatibility — a sprawling category that encompasses everything from e-commerce platforms to labor markets that match employees with employers.

Kerimov and his colleagues focused on programs that match live kidney donors with people who need transplants. Live donors typically volunteer to give one of their kidneys to a loved one. But biological differences between a donor and their intended recipient can render the pair incompatible.

Kidney exchange programs solve this problem by swapping donors amongst different patient-donor pairs, choreographing a kind of kidney-transplant square dance aimed at finding a compatible partner for every willing donor.

In countries such as Canada and the Netherlands, kidney-matching programs perform a batch of matches every few months (called periodic policies). American programs, meanwhile, tend to perform daily matches (called greedy policies). Both models seek to produce the greatest number of high-quality transplants possible, but they each have advantages and disadvantages.

Less frequent matches in a periodic policy allow more patient-donor pairs to accumulate in the kidney exchange network, creating potential for better matches over time. But this approach risks making some patients sicker as they wait for a better match that might never appear.

Arranging feasible matches as soon as they become available in a greedy policy avoids that predicament. But it means passing up the opportunity to make a potentially better match that could represent the possibility of a longer, healthier life.

Balancing these trade-offs is tricky. There is no way of predicting precisely when a patient-donor pair with a particular set of characteristics will show up at the kidney-exchange network. And in the world of organ transplants, there are no do-overs.

Kerimov and his colleagues have constructed a mathematical model that represents a simplified version of a kidney exchange network.

Within the model, the researchers could dictate which patient-donor pairs could be matched with one another. They can also assign different values to individual matches based on the number of life years they provide. And they can establish the probability that various kinds of patient-donor pairs with particular characteristics might arrive at the network and queue up for a transplant at any given time.

Having set those parameters, the researchers applied different matching policies and compared the results. As it turns out, the answer to whether one should wait or not is: It depends.

To determine which policies generated the best outcomes — i.e., performing matches either daily or periodically — the researchers calculated the difference between the total value in life years that could possibly be generated within the network and the amount generated by a specific policy at a particular point in time. The goal was to keep that number, evocatively dubbed "all-time regret," as small as possible over both the short and long term.

In their first paper, Kerimov and his team explored a complex network in which donor kidneys could be swapped amongst three or more patient-donor pairs. When such multiway matches were possible, the cost of applying a daily-match policy turned out to be onerous. Using all available matches as quickly as possible eliminated the chance of later performing potentially higher-value matches.

Instead, the researchers found they could minimize regret by applying a periodic policy that required waiting for a certain number of patient-donor pairs to arrive before attempting to match them. The model even allowed the team to calculate precisely how long to wait between matchmaking sessions to get the best possible results.

In their second paper, however, the team looked at a simpler network in which kidneys could only be swapped between two donor-patient pairs. Here, their findings contradicted the first: Applying a daily-match policy minimized regret; a periodic matching process yielded no benefit whatsoever.

To their surprise, the researchers discovered they could design a foolproof algorithm for making two-way matches in simple networks. The algorithm employed a ranked list of possible match types; and the researchers found that no matter how many patient-donor pairs of various kinds randomly arrived at the network, the best choice was always simply to perform the highest-ranked match on the list.

In future research, Kerimov hopes to refine the model by feeding it data on real patient-donor pairs that have participated in actual kidney exchange programs. This would allow him to create a more realistic network, more accurately calculate the likelihood that particular kinds of patient-donor pairs will show up, and assign values to matches based not only on life years but also on rarity and difficulty. (Certain blood types and antibody profiles, for example, are rarer or more difficult to match than others.)

But Kerimov already suspects that in a real-world situation, the wisest course of action will be to alternate between periodic and greedy policies as circumstances dictate. In a simple region within a kidney exchange network that only allows for two-way matches, pursuing a greedy policy that involves taking the first match that appears on a fixed menu of options would be the best choice. In a more complex region that allows three-way matches, however, pursuing a periodic matching policy that involves waiting to make rarer and more difficult matches would ultimately offer more patients more years of healthy life.

The benefits of choosing flexibly between greedy and periodic policies should hold for any kind of matching market that can be represented by a network with simpler and more complex regions, such as a logistics system that matches online orders to delivery trucks or a carpooling system that matches passengers with drivers across different parts of a city.


This article originally ran on Rice Business Wisdom and was based on research from Süleyman Kerimov, an assistant professor of management – operations management in the Jones Graduate School of Business at Rice University.

Trending News



A research team housed out of the newly launched Rice Biotech Launch Pad received funding to scale tech that could slash cancer deaths in half. Photo via Rice University

A research funding agency has deployed capital into a team at Rice University that's working to develop a technology that could cut cancer-related deaths in half.

Rice researchers received $45 million from the National Institutes of Health's Advanced Research Projects Agency for Health, or ARPA-H, to scale up development of a sense-and-respond implant technology. Rice bioengineer Omid Veiseh leads the team developing the technology as principal investigator.

“Instead of tethering patients to hospital beds, IV bags and external monitors, we’ll use a minimally invasive procedure to implant a small device that continuously monitors their cancer and adjusts their immunotherapy dose in real time,” he says in a news release. “This kind of ‘closed-loop therapy’ has been used for managing diabetes, where you have a glucose monitor that continuously talks to an insulin pump. But for cancer immunotherapy, it’s revolutionary.”

Joining Veiseh on the 19-person research project named THOR, which stands for “targeted hybrid oncotherapeutic regulation,” is Amir Jazaeri, co-PI and professor of gynecologic oncology at the University of Texas MD Anderson Cancer Center. The device they are developing is called HAMMR, or hybrid advanced molecular manufacturing regulator.

“Cancer cells are continually evolving and adapting to therapy. However, currently available diagnostic tools, including radiologic tests, blood assays and biopsies, provide very infrequent and limited snapshots of this dynamic process," Jazaeri adds. "As a result, today’s therapies treat cancer as if it were a static disease. We believe THOR could transform the status quo by providing real-time data from the tumor environment that can in turn guide more effective and tumor-informed novel therapies.”

With a national team of engineers, physicians, and experts across synthetic biology, materials science, immunology, oncology, and more, the team will receive its funding through the Rice Biotech Launch Pad, a newly launched initiative led by Veiseh that exists to help life-saving medical innovation scale quickly.

"Rice is proud to be the recipient of the second major funding award from the ARPA-H, a new funding agency established last year to support research that catalyzes health breakthroughs," Rice President Reginald DesRoches says. "The research Rice bioengineer Omid Veiseh is doing in leading this team is truly groundbreaking and could potentially save hundreds of thousands of lives each year. This is the type of research that makes a significant impact on the world.”

The initial focus of the technology will be on ovarian cancer, and this funding agreement includes a first-phase clinical trial of HAMMR for the treatment of recurrent ovarian cancer that's expected to take place in the fourth year of THOR’s multi-year project.

“The technology is broadly applicable for peritoneal cancers that affect the pancreas, liver, lungs and other organs,” Veiseh says. “The first clinical trial will focus on refractory recurrent ovarian cancer, and the benefit of that is that we have an ongoing trial for ovarian cancer with our encapsulated cytokine ‘drug factory’ technology. We'll be able to build on that experience. We have already demonstrated a unique model to go from concept to clinical trial within five years, and HAMMR is the next iteration of that approach.”

Trending News