Nishiura and son
What are you currently researching?
My research interests span the areas of statistical epidemiology of infectious diseases, epidemic modeling and biomathematical formulation of infectious disease spread. A common thread in my research is in understanding the underlying epidemiological dynamics behind empirically observed data. Having a background in medicine and being a licensed medical doctor in Japan, I became an epidemiological modeler because I strongly believe that population science is crucial for elucidating the most effective strategy for infectious disease control. Among all subjects, I’m presently addicted to spatiotemporal prediction of Zika virus transmission. Of course it is attractive for everyone to study the Zika virus spread in 2016 through Rio de Janeiro Olympic games, but it is more scientifically enchanting to consider if the virus can persist to widely infect a substantial fraction of people across the world, to understand how large the epidemiological impact of the epidemic (e.g. global burden of Zika) would be, and to quantitatively capture the impact of global warming on its transmission in longer time spans.
What initially attracted you to epidemiology?
When I was a med student, I was interested in global health and international cooperation, and I had a few opportunities to be exposed to immunization practices in remote settings in South and Southeast Asia through internships at governmental and non-governmental organizations. Before the internship,my impression toward the immunization program was a somewhat slowly progressing “boring” one, compared with much more dynamic experiences such as surgical operations of wounded soldiers in refugee camps. However, I was totally convinced that I was wrong. I came to know that scientific grounds of the immunization program were supported by rich population theories and solid mathematical formulas. I was lucky to have had a good supervisor there, because I was then instructed to read “Infectious Diseases of Humans” written by Anderson and May (1990). The mathematical complexity in that book was enough to give up becoming a clinician.
Have you encountered any surprises in your research?
In all of my quantitative studies, I have seen that statistical likelihood function does not swiftly lead to convergence whenever my formulation is wrong, even slightly. Sometimes the error can be simple coding error. Or just a mathematical mistake in my derivation. However, most frequently, the reason why I did not come up with nice convergence has been the error in my mathematical formulation. Every single time I’m correct in formulating an existing problem, I’m shocked to find that the right value can be found immediately. The speed of statistical estimation studies is always determined by our capacity to correctly capture existing phenomena laying in front of us.
Do you ever find the complexity of biology daunting?
As I work on epidemic modeling with increased realism, lately I observe that everything is too complex. Nevertheless, very simple rules or metrics can also be found from a complex system. For instance, the global spread of emerging epidemic disease has been known to be induced by the complex airline transportation network. While the network itself has been really complex, the so-called “effective distance” invented by Brockmann and Helbing (Science, 2013) has excellently shown that the use of shortest path and its weight can predict the time at which the disease is imported. We cannot capture all realistic dynamics using a simple system, but sometimes an extraordinary simple description can explain substantially large part of seemingly complex system at a glance. Finding and using such rules or metrics are the most exciting moment.
What are the next big breakthroughs in your field?
A prediction system using both microscopic and macroscopic data lies as the next challenging step in epidemiological modeling. Not only merging theoretical approaches in epidemic modeling with those in population genetics, but also employing other useful biological theories remains to be a big challenge in epidemiology. For instance, we have yet to consider how we simultaneously handle models in systems biology and epidemiology. Also, without fully clarifying functional roles, we are motivated to use omics approaches in epidemic modeling. By doing so, predictable models could newly arise from previously completely unpredictable matters.
What aspect of your research do you find most intriguing?
Mathematical formulation of infection process is the most favorite part of my job. My formulation is intended to estimate key parameters from empirical data, and that approach is quite different from mathematical studies with existing compartmental systems. To capture realistic features of epidemic events in line with the data generating process, the formulation should allow us to maximally quantify the system based on minimally available datasets.
What are the most significant challenges you’ve faced as an interdisciplinary scientist?
We have to speak several different languages in our daily life. My team members include mathematicians, statisticians, biologists, epidemiologists and informaticians. All of them speak very different languages at the same time, but that chaotic situation represents why we love to survive in this cutting edge field. Always we face dilemma of cutting edge studies, e.g. mathematicians work on a fixed specific model, but it later turns out to be biologically implausible. Sometimes, that might be stressful, but I believe that interdisciplinary researchers must enjoy that kind of life in our complex system!
What would your message to a young and aspiring mathematical biologist be?
Please try not to perfectly capture natural phenomena in equations. No model can replace reality. Nor please do not pretend to have captured all the essence using a simple model. Our model in tractable equations is always just a caricature of reality. Among all mathematical models, epidemiological models are the worst: they can never be true. Good models always have to find somewhere between as an optimal compromised solution.
How did you get introduced to mathematical biology?
When I started my PhD career, my promoter, Masayuki Kakehashi advised me to read two books. One was written by James Murray on mathematical biology (Springer, 2002), while the other was by Yoh Iwasa written in Japanese (“Introduction to Mathematical Biology”, Kyoritsu, 1990). These two were not too difficult and perfectly fit me with a medical background. I would stress that Yoh’s book has changed my brain, because I did not have any background of stability analysis in advance of reading his book and his guide to understand model behaviors has induced my revolution in the theoretical understanding of life. Also, I had been new to stochastic modeling, but his book brought me to even seriously derive a likelihood function from an explicit stochastic model by differentiating the prob-ability generating function. I do hope that Yoh will translate his into English.
What is the best professional advice you have ever received?
“Work with a global leader. The leader knows how to reproduce next leaders. Others would seldom know.” That has turned out to be true.
If you were not a scientist, what would you be?
I would have been a gemologist. Bringing my kids to mountains, I always love differentiating different types of stone there. Stone tells us the history of earth. Moreover, the professional capacity to act as a gem identifier sounds very exciting to me: the expertise sounds very special!
If you have any spare time, what do you do when you are not working?
I love jogging. Whenever I find time, I bring my family, friends and team members to run together. Every single year, my team members and I enter a relay marathon under the sun in the middle of summer season. Rather than drinking, doing our best in that marathon event is the way to make friends in my department.
Dr. Hiroshi Nishiura is Professor in the Department of Hygiene at the Graduate School of Medicine, Hokkaido University. For more info, visit: http: //plaza. umin. ac. jp/ ~infepi/ hnishiura. htm