Interview with Dr. Naoru Koizumi

Dr. Naoru Koizumi

Today we're joined by Dr. Naoru Koizumi, a professor of public policy and the associate dean of research and grants at the Schar School of Policy and Government at George Mason University.

Dr. Koizumi specializes in medical policies related to organ transplantation, chronic diseases, and healthcare optimization. Her research focuses on applying mathematical modeling, data analytics, and GIS to complex healthcare issues.

She's also deeply involved in mentoring high school students in the Biology and Medical Data Analytics Internship Program, where she helps students engage with cutting-edge research in biomedical science. Recently, she’s joined the NSF as well—congratulations on that!

Isaac

Thank you so much for joining us today, Dr. Koizumi. It’s an honor to have you here to share your insights. To start off, I’d love to learn more about your journey. Could you tell me what inspired you to combine math with healthcare policy, especially in areas like organ transplantation and chronic diseases?

Dr. Koizumi

My academic background is in regional science. I received my PhD in regional science in 2003 from the University of Pennsylvania. Regional science is a multidisciplinary, methodological field. We’re exposed to techniques like optimization and dynamic programming, along with GIS. These tools are usually applied to urban issues—such as how cities grow or how companies cluster in specific areas.

However, while I was studying this, I also worked as a statistician at Penn’s medical school and hospital in the psychiatric department. That’s where I first saw the potential to apply optimization, math, and statistics to healthcare.

As I started to apply those tools in healthcare, I realized I was much more interested in that field than in urban and regional issues. So, I pursued a second PhD in medical science, although I did it at a Japanese university rather than at Penn, which had more stringent requirements.

Isaac

Could you explain how Geographic Information Systems (GIS) are used in your work to understand healthcare issues like organ distribution, diseases, or even urban development?

Dr. Koizumi

In my current work, I primarily use GIS and spatial analysis to visualize conditions or research questions. For example, in organ allocation, the demand for transplants and the supply of organs differ significantly across the United States. Simply stating this is one thing, but if you visualize it on a map, it’s much more powerful. For instance, patients waiting for transplants in California face significant challenges compared to those in other states.

However, spatial statistics, such as spatial regression, can be complex because typical regression methods can introduce biases when dealing with spatial and temporal data. The challenge is correcting for these biases. If you’re just a statistician, you may not know how to handle that. But if you’re trained in GIS and regional science, you learn how to address these challenges in more accurate ways.

Isaac

Along with GIS, you also use mathematical modeling techniques like optimization and simulations. How do these models help improve outcomes in organ transplantation and disease treatment?

Dr. Koizumi

As I work in a policy school, all of my research is tied to policy formulation. Organ transplantation is deeply linked to policy, especially when it comes to organ allocation.

As I mentioned earlier, where you live can significantly impact how quickly you receive an organ. Currently, there are geographical boundaries determining who has access to organs in certain areas. But these policies are changing to create a more equitable system. The question is: How do you adjust these boundaries to make the system fairer? Mathematical models help answer these questions, and I use my research to inform transplant communities—surgeons, center administrators, and government officials—about optimal allocation boundaries.

Although I’m not a policymaker, I can use my research to suggest better boundaries and highlight the consequences of not having an optimal system.

Isaac

When deciding organ allocation, how do you balance the priority of waiting times with other factors, like the likelihood of success, in order to ensure a fair distribution?

Dr. Koizumi

That balance is always changing. It’s a complex issue involving human behavior. Some factors, like patient behavior, are already incorporated into allocation systems, but there are still many factors that aren’t.

For example, patients in poorer areas, or African Americans in underserved communities, may not have access to good nephrologists, who are the ones who decide whether a dialysis patient is referred for a transplant. This inequality has been an issue, but in 2020, policies changed to account for dialysis time when determining transplant eligibility.

Now, if you’ve been on dialysis for five years and finally get listed for a transplant, your wait time starts from when you were first diagnosed, rather than from when you were officially listed. This policy change is one example of how researchers’ empirical studies can inform policy changes that make the system more equitable.

Isaac

One of your projects, funded by the Bill and Melinda Gates Foundation, used ICT tools to improve medical adherence for tuberculosis patients in India. How did advanced technologies like machine learning and information and communication technologies (ICT) improve public health in this project?

Dr. Koizumi

This project took place in the slums of Ahmedabad, India. In these communities, antimicrobial-resistant tuberculosis was on the rise due to poor adherence to medication. The question we aimed to address was: How can we improve adherence to antibiotics?

We had a simple idea: Send text message reminders to people in the slums. Everyone had cell phones, so this seemed like a feasible way to get the message across.

In addition to text reminders, we also provided healthcare visits, explaining the importance of taking medication and encouraging patients to visit treatment centers. But what we found in our study was that the personality of the healthcare staff at the centers had a huge impact.

If the staff were enthusiastic, energetic, and genuinely cared about the patients, it made a much bigger difference than any technology.

My lesson from this project is that human factors must be combined with ICT. A message that comes from nowhere, without any personal connection, doesn’t have the same impact as one that comes from a trusted individual. ICT alone won’t achieve what we want, but when it’s combined with human touch, it can be powerful.

Isaac

You’ve also studied the issue of illicit kidney trade. Can you share your insights on that?

Dr. Koizumi

Yes, this issue is often referred to as organ trafficking. It’s a serious global problem, and we’re seeing increasing awareness of it in many regions.

Isaac

Can you share how mathematical analysis can help address a complex problem such as kidney organ trafficking?

Dr. Koizumi

So there are a lot of field researchers—well, I shouldn’t say a lot, but there are many so-called qualitative researchers. These are well-known people who go to slums, and also refugee camps in Turkey and Egypt, and they interview people. For example, in Egypt, refugees may sell their kidneys to get a passage to Europe. In poorer areas of developing countries, people sell kidneys just to escape poverty and have a better life.

This is called kidney or organ trafficking in general. Some people call it transplant tourism.

What’s interesting to me is that I have a clinical background in kidney and liver diseases, and also a quantitative background. When I observe this phenomenon, it’s like a whack-a-mole game. For example, China was once a hotspot for organ trafficking, but then they tightened up their policies, and China became less well-known for it. As a result, trafficking shifted to other countries. Those countries then started applying more stringent regulations, and the trafficking shifted again.

So these hotspots change dynamically, based on not just policies, but income levels, and the relationships between countries. The hotspots that were known 20 years ago are completely different from those that were prominent 10 years ago, and today we have an entirely different set of countries.

From a policy point of view, how can we regulate or control these organ sales better? It's like a whack-a-mole game—you can't just apply sanctions or try to control things once they become well-known hotspots. We need to predict and understand the underlying factors that make a country become a hotspot for organ sales.

That’s where mathematical modeling comes in. We can develop prediction models and tease out the factors that make certain countries become hotspots for organ trafficking.

Isaac

Dr. Koizumi, you've mentored many high school students through the GMU Biomedical, Biology, and Medical Data Analytics Internship Program. What do you find most rewarding about working with students at this level, and how do you help them navigate complex tasks and issues in biomedical research?

Dr. Koizumi

We’re very selective in choosing internship students, and we’ve been amazed by the quality of high school students we get every summer. I have to say, the quality is top-notch. We’re not just getting the usual high school students.

It’s been a pleasure working with these highly qualified students, and it’s especially rewarding when I hear that a student continues to do research in areas like data analytics in biology. One student, for example, took a bio analytics course two years ago, and after working with me for a year, she ended up at the University of North Carolina, majoring in statistics.

That’s the most rewarding part—to hear that you’ve motivated someone to pursue a career in the field.

Isaac

What advice would you give to high school students interested in pursuing a career in statistics or biomedical research, and how can they best prepare to succeed in these fields?

Dr. Koizumi

I do a lot of multidisciplinary research, and that’s key to my work. For example, I can’t develop optimization models on my own, nor can I create new models in operations research (OR). I don’t know much about it, but I know enough to understand what OR can do and where the latest technologies and models are. I work with mathematicians who specialize in modeling to do research.

In data analytics, one challenge is that some data analysts aren’t interested in learning about the fields they’re analyzing. For instance, if a data analyst isn’t interested in kidney disease or organ trafficking, that limits their ability to do something novel.

So, my advice for students interested in data analysis is to stay open to learning about other fields. Whether it’s biology, organ trafficking, human trafficking, or environmental issues—understanding those areas is crucial. Data analysts need to broaden their perspectives and be willing to collaborate with experts in other domains.

Multidisciplinary research is increasingly important, and agencies like the NSF and NIH emphasize this. They used to fund small-scale studies by individual researchers, but now they focus on larger, multidisciplinary studies. Being open to different fields as a data analyst is critical in today’s research environment.

Isaac

Looking ahead, how do you see the role of mathematical modeling evolving in healthcare, and what do you think the future holds for this multidisciplinary, interdisciplinary field?

Dr. Koizumi

Mathematics has always been helpful, but today everyone’s talking about AI. AI has a lot of potential and implications, but there are also a lot of challenges. One of the biggest concerns is making sure AI results are interpretable and that the solutions provided are fair. Understanding what fairness and interpretability mean is a complex issue, and that’s where interdisciplinary collaboration is key.

AI experts need to work with domain experts, ethicists, and policy professionals. This collaboration is essential to ensure AI solutions are used effectively and ethically in healthcare.

Isaac

For the last question, what has been the most rewarding part of your research, particularly in using math to solve real-world health problems?

Dr. Koizumi

For me, the most rewarding part is collaboration. I have a large team, and finding good collaborators is always difficult. I’ve worked with qualitative researchers who don’t always understand or appreciate quantitative analysis, and it takes time to build mutual understanding.

Physicians can be similarly challenging—some are very interested in AI and modeling, while others aren’t even open to working with statistical analysis. It’s a process to develop good collaboration, but I’ve been fortunate enough to find amazing collaborators. Without them, I wouldn’t be able to do the interesting research that I do.

Now that I’m going to the NSF, my goal is to learn what good research looks like and contribute to funding high-quality projects.