Mi-Ok Kim, PhD, joined UCSF in 2016 from the University of Cincinnati, where she worked on high-profile research into pediatric diseases with doctors and researchers at Cincinnati Children’s Hospital. In several cases, her statistical contributions became methods that now form part of a standard menu of options, such as semi-parametric inference for causal effects, historical borrowing adaptive trial designs for pediatric patients, and random coefficients model analysis of repeated measures.
Kim has held leadership roles at UCSF since she arrived – since 2016, she has been director of the biostatistics shared resource core at the Helen Diller Comprehensive Cancer Center. In 2020, she added a role as chief of the division of biostatistics when Chuck McCulloch, PhD, became vice chair of operations. Now, Kim is adding the role of vice chair of finance.
We spoke to Kim to learn about her approach to leadership and biostatistics. We found her funny, modest and yet really passionate about her work. Excerpts from the interview (lightly edited for clarity) follow.
What will you be doing as vice chair of finance?
As a part of the leadership team, I will contribute to the vision of the department and its growth and specifically assist the chair from the financial aspect, ensuring financial integrity and accountability of the department budget.
Will you bring any statistical wizardry to the role?
Oh, I think it might be quite the contrary because statistics is all about probability and conceptualizes constructs as random events. Minimizing randomness and fluctuation in the budget by pre-planning would be a goal.
What has made you raise your hand for leadership roles?
I didn't really think much about it, perhaps because I started out on a conventional arts and sciences campus where it was more egalitarian. People took turns being the department chair. Chuck McCulloch was such a big figure and a character, that when he stepped down, I was just trying to step into his shoes – and then trying to fill them as much as possible. For the efficiency of the unit, I am making decisions, but, after all, I'm their colleague.
The part that really interests me is this: A lot of the senior faculty in the biostatistics division, including Chuck, are retiring. So, we are at the point where recruitment is essential for us to continue to be productive and successful. I wanted to contribute to that, and I also want to contribute to mentoring the junior faculty, knowing what I know.
As for the Vice Chair of Finance position, I wanted to use my experience managing finances on the Cancer Center side. I have been managing the biostatistics shared resource core budget, so I'm already familiar with the financial aspect of this business – which we hope is a good thing for the department.
Isn’t it a lot to lead a division and be vice chair?
Statisticians are very reasonable, so there’s not much drama. Of course, there’s also the joke that if you put three statisticians in a room, you’re going to hear five opinions.
How did you get into statistics from an undergraduate degree in English Education at Pusan National University in Korea? And to what extent have being a woman and an immigrant been obstacles to overcome?
Throughout my career, I was asked about how much of my perspective is shaped by being a woman. I’ve never really felt that there’s any real limitation. I grew up in the time period that Korean women were moving into areas conventionally reserved for men, similar to what happened in America in the 1970s and 1980s. My generation felt we had as much chance as men to make difference in the world.
What affected my life and career more is me coming to the U.S. I studied English because my father always wanted one of his three girls to be an English teacher. I came to the U.S. to study English education. It was a huge mistake! I found myself taking math courses as electives. No non-math majors take math courses as electives!
In an education measurement course, we had a guest lecturer who was a statistician. After the lecture, he offered me a chance to switch over into statistics. That’s how it happened! He didn't know much about me, but he said, “Oh, you’re Asian, you must be good at math!” So, I actually benefitted from that stereotype.
How did you move from statistics to biostatistics?
You know, statistics is a lot like law: It applies to everything, you just have to find which type you want to specialize in. For me, there was no noble calling that pulled me to biostatistics. When I was teaching statistics [at the University of Kentucky], I was commuting. Then, with my second child, I wanted to cut down on the commute, and we had hospitals nearby.
I wish I had a noble mission or cause. But I like biostatistics because I'm the kind of person who truly blossoms in interdisciplinary and collaborative research. I know that I'm good at what I bring to the table. I'm expecting others to bring their own expertise, so I don't need to be the master of everything.
Often, biostatisticians end up being middle authors on publications, even when they’ve made big contributions. Can you tell me about some of the publications you are particularly proud of?
You’ll want to find it the one manuscript that I'm the co-first author on! It's on pediatric Crohn’s disease. The anti-TNF alpha treatment had just been introduced for adults, and they wanted to see whether it would also benefit children. A randomized trial is hard and takes a long time. So we looked at data from an observational study of off-label use on children 17 and under to try to obtain causal effects like a randomized trial. Nowadays, causal inference has become a very common tactic, but it was the first causal inference work in pediatrics. The statistical treatment of confounding was essential to the study results, so I got to be the co-first author. The beneficial effect of anti-TNF was shown, and now it is included in the standard treatment regimen for pediatric Crohn’s patients.
The other paper was on neurofibromatosis type 1 [a condition where benign tumors develop on nerves]. There were 2 consortia: The pre-clinical consortium conducted all kinds of mouse trials to identify which compounds were promising; those components moved forward to the clinical consortium. All that work led to that one FDA-approved drug.
Back then, this approach didn’t have a name, but now it’s called a platform-type design. Well, like a human, each mouse is different. The tumors don’t grow very uniformly. So we wanted some way to moderate that heterogeneity. And we were wasting a lot of mice having equal-sized control mice every time.
What the platform trial design allows is pooling control groups serially as long as we ensure no genetic drift in the control group and nothing strange happens. So we do not need to have an equal-sized control group every time; we can combine the control mice and then use that group to compare with different intervention groups.
So it’s more humane?
It’s cheaper and more efficient. I think the job of the statistician in biomedical research is to bring logic and reasoning. Biostatistics is the science of how to conduct biomedical research. We bring control and rigor to the process. We are building the systems to produce valid and reproducible research.
I love what I do, as you can see.