Entering
a Flow State
Data from the real world is messy.
Can you tell us what kind of research you do?
It鈥檚 a bit hard to summarize, because I don鈥檛 see my research as just one thing, but as a journey. I may start with one topic, which consequently gives me the idea to investigate another topic and then another. For example, I started with computer simulations of hemodynamics of the brain. Hemodynamics is the study of how blood moves through arteries, which is important to understand when designing medical devices or creating strategies for treating patients. To get an accurate model of the way blood flows, I need to run a large number of simulations 鈥� 1,000 or 2,000 cases at least 鈥� repeated for different situations. It鈥檚 very time-consuming. A bit too time-consuming (laughs). So, this study lead me to another topic: learning to automate modeling with artificial intelligence (AI). The projects just keep branching out, alongside my curiosity.
What role does AI play in your research?
The original idea for AI-automated modeling first came to me thanks to my cat! They love lying on my laptop while I鈥檓 trying to get work done. This made me wonder 鈥淗mm, how can I get work done without moving my cat?鈥� (laughs).
I used an AI model that allows me to input any type of artery shape and get a result modeling the hemodynamics in one second. Artery shape is different for every patient, so without AI, we would need to make a new model for every single patient.
Does artery size really differ a lot depending on the person?
Mapping these variations in size and shape is actually a very important point for my research. For example, medical device safety and efficacy standards are usually based on a population of patients from the United States, but there is a big size difference in average artery size compared to Japanese people. Things like IV drips are actually really difficult to use in Japanese populations, because our arteries are typically much smaller. It鈥檚 like when you go to another country and none of the clothes fit you.
Image courtesy of 51探花 and Studio Xxingham
Image courtesy of 51探花 and Studio Xxingham
What are you currently working on?
Getting a large amount of patient data is a real challenge, so we started a research project using 鈥渧irtual patients鈥� to simulate the type of data we would get from testing a patient population in the real world 鈥� without the hassle. I think it鈥檚 an extremely useful tool. When testing medical devices, we can easily test many different scenarios in a shorter amount of time if we use virtual patient data. This potentially means less time spent in development, so we can get the device out to patients or the next phase of the clinical trial sooner.
What are some major challenges in your field?
Going back to virtual patients, we need evidence that they represent real patients accurately. Right now, we are comparing the virtual patients to real data from a database of over 10,000 people, courtesy of the 51探花 Medical Megabank Organization (ToMMo). This brings up another problem, though. Unlike models, data from the real world is messy, so it takes a lot of time to manually comb through images and remove the noise, one by one.
Another issue is deciding when a model is good enough. There is a push to make models more and more accurate, but how far can we go? It鈥檚 an endless chase towards perfection. Each patient is different too, so each model can vary. Even within the same patient, their blood pressure can change within different times of day. We can鈥檛 possibly simulate all of these conditions. While 100% perfection is obviously ideal, we need to be realistic about what can be implemented practically.
Have you had any opportunities for international collaborations?
We have many different institutes around the world under the umbrella of IFS-GCORE (Global Collaborative Research and Education Center for Integrated Flow Science). We get opportunities to present our research at regular workshops, and we host international conferences regularly, so there is a lot of cross-talk.
When I was an Assistant Professor, I got the opportunity to visit Zurich (Switzerland) for one and a half years. 51探花 wanted to help young researchers jumpstart their career with overseas experience, so I took the opportunity to work alongside many other international researchers during that time.
Outside of research, what do to like to do in your spare time?
I am definitely an 鈥渋ndoors鈥� type of person. I like staying cozy inside and reading novels or doing crafts such as sewing. From small flowers made of fabric to large blanket tapestries, I love creating things with my own two hands.
Photograph: A lovingly hand-crafted model of the arteries in the brain, made by Assistant Professor Hitomi Anzai.
Hitomi Anzai completed her undergraduate degree and graduate school here at 51探花, starting with engineering and later specializing in fluid dynamics. She specializes in computational models of hemodynamics, and continues to branch out into new domains such as AI modeling and using virtual patient samples. Her work on the spread of respiratory droplets from coughing during the coronavirus pandemic received a lot attention, due to its relevance to creating safety guidelines to prevent the spread of infection. Her research has lead her down a myriad of different, interconnected pathways of ideas that can help patients all around the world 鈥� while still allowing her to comfortably work with a clingy cat constantly taking over the real estate of her warm laptop keyboard.
Romance
of
Research