Breadcrumb navigation

Message for students 2025:Daisaku Shibata

March 19, 2025

Aiming for a natural language processing AI for wide use in medical establishments

Daisaku Shibata
Assistant Manager
Biometrics Research Laboratories

After completing a doctorate program with research on clinical natural language processing , Shibata worked for two years as a specially appointed assistant professor at university. Later, he joined NEC in April 2023. Engaging in the research of clinical natural language , which is his specialty, he contributed to the commercialization of an electronic medical records system using a large language model. He is involved in working style reformation in medical settings by means of streamlining medical documentation. He is actively submitting research papers on findings from this effort and participating in academic conferences.

From academia to the industrial world

After working at a graduate school for two years as a specially appointed assistant professor, I joined NEC as a mid-career employee in 2023. My reason for transferring from academia to a company is because I also wanted to work on clinical natural language processing from the industrial side. I wanted to know how I can do research and development at a company, so I considered changing jobs.

The reason for choosing NEC over many other companies is this talk with an NEC employee that I met at an academic conference. He mentioned that NEC is also working on clinical natural language processing, which piqued my interest. In fact, I am still studying clinical natural language processing by carrying on my research theme from before joining NEC.

The difference with academic research is that now I need to consider the balance between research and practical application. For example, when making a system for use at hospitals, using a large model may be difficult due to its massive consumption of computational resources. In such cases, optimization is required while considering the overall costs. Such juggling management may be something unique to research and development conducted by an enterprise.

NEC is also a vendor of electronic medical records system delivered to hospitals. This is very advantageous for engaging in R&D of clinical natural language processing. We can work toward commercialization while linking with a medical records system actually in use at hospitals. In fact, the electronic medical records system for which I participated in the R&D was marketed in April 2024 as MegaOak/iS. This is a system that assists medical documentation using large language model s. Being able to develop a system that is actually put to use by many people in medical settings is one of the rewarding aspects of working at NEC.

Nevertheless, commercialization is not the end of the line of R&D. I believe it’s important to continue on the cycle of production where a product is actually put to use and we receive feedback on the product to make improvements.

No huge difference with research in academia

Clinical natural language processing is a field of research where we apply natural language processing to text and audio data generated in medical settings for the purpose of assisting documentation and diagnoses. I generally work with text data. Medical documents are technical and not written in plain language. Wording used in medical records is non-grammatical and often is a series of information segments. For this reason, we need to create a dedicated model. Medical knowledge and know-how are also necessary, so we at NEC collaborate with Tohoku University Hospital and other universities and medical institutions for R&D. Having access to data created at the actual sites is a blessed environment in terms of machine learning R&D.

This may sound like we are only working on commercialization or applied research on site, but we are actually still writing research papers and partaking in academic conferences in parallel. It has almost been two years since I joined NEC, and during this time, I have made a presentation at a conference. Another presentation is scheduled for the coming March. I want to emphasize that you CAN continue academic activities even as a corporate worker. You also have opportunities to engage with laboratories that you were affiliated with in graduate school. Thanks to this situation, I am able to work on research without feeling much gap between here and the academia.

Develop your unique strength

One of the positive qualities of NEC is that it’s up to you whether you want to lean toward research or be more focused on development in terms of R&D. It is nice that researchers can adjust that balance according to their own specialization and preference. Taking advantage of this environment, I hope that you can be a researcher who can establish your distinctive style. A researcher who can put a face on their research theme is powerful. I am trying to find my uniqueness in the R&D of clinical natural language processing. As students, you may want to establish your own strong point that leaves you unrivaled.

At the same time, you do not need to be confined to the research theme you are currently working on. Even in my team, not many have studied clinical natural language processing as their specialty in the past. The majority are researchers who have been working on other research themes. Many are active in companies, exploiting new areas by leveraging their knowledge from learning and their know-how from research. NEC’s range of business covers a spectrum of research themes from undersea to space. I hope that you can find and work on a theme that you can actively engage in.

  • *
    The information posted on this website is the information at the time of publication.

A day at work

Message to my past self in my school days

Private column

I recently started working out. I used to play basketball during my schooldays. I haven't been able to exercise for the past few years, and I am starting to lose physical strength (laughs). I am currently working out and jogging to catch a breath in between work while trying to get my strength back.