お問い合わせ | 化学のものづくりに特化したデータ解析・機械学習クラウドサービス

Datachemical, Inc.

Datachemical LAB Case Study Interview

From the research and education fields

Professor, Chemistry and Life Science Program, Faculty of Engineering, University of Miyazaki

Data Chemical Co., Ltd. Representative Director and CEO

"No-code x Materials Informatics" spreading to research and education

Materials informatics (MI), which is essential for accelerating material development, is now spreading not only in industry but also in education.
However, MI education requires high hurdles, such as specialized knowledge and programming (such as Python), and introducing it into the curriculum is not easy.
In this interview, Professor Tatsuya Oshima of the Faculty of Engineering at the University of Miyazaki will discuss a case in which he used the no-code MI tool "Datachemical LAB" in his own laboratory and university lectures, allowing students to "set their own problems and verify hypotheses using AI."
Practical education that develops the ability to "think with data" - what is the background, real results, and outlook for the future? Be sure to take a look at the cutting edge of digital transformation in education.

Guest

Prof. Tatsuya Oshima —
Professor, Chemical Life Program, Faculty of Engineering, University of Miyazaki

Interviewer

Shogo Yoshimaru —
CEO, Datachemical Inc.

Yoshimaru (Moderator):

Thank you for your time today. It has been about three years since your university adopted Datachemical LAB. To start, could you tell us how you have been using it in your own research?

Prof. Oshima:

My research focuses on solvent extraction of metal chloride complexes. Conventionally, attention in this field tends to center on the design of metal complexes themselves. However, I realized that the physical properties of the solvents themselves are critical parameters in metal extraction, which led me to start collecting and organizing solvent data.
I initially used solubility parameters to qualitatively organize extraction capacity, but I thought there might be room to go further — and began exploring whether I could quantitatively predict extraction rates.

Yoshimaru:

And that's when you turned to machine learning?

Prof. Oshima:

Exactly. At the time, the assumption was that you had to write code in Python. Since our lab — myself included — doesn't have a background in computer science, sustaining that approach was genuinely difficult. Then Datachemical LAB was released, and I thought, "This we can work with." From there, machine learning adoption in our lab expanded rapidly.

Yoshimaru:

How did things change after you introduced Datachemical LAB?

Prof. Oshima:

The biggest difference was simply that no programming is required. That removed an enormous barrier for students. For chemistry students, learning Python from scratch just to build a model is a very high hurdle. But with Datachemical LAB, no prior background is needed.
As a result, going back to Python is no longer even an option for us. Our policy is: if we're doing machine learning, we use Datachemical LAB.

Yoshimaru:

Has this translated into tangible results?

Prof. Oshima:

We started with an extraction rate prediction model that achieved an R² of around 0.94. Since then, the volume of data students handle has grown and our validation methods have evolved. We are now building models with R² values of 0.98 to 0.99 — quite high accuracy.

Yoshimaru:

Your university has signed a partnership agreement with us with the aim of collaborating on education. As part of that, this year you actually used Datachemical LAB in the classroom.

Prof. Oshima:

Yes. The Faculty of Engineering at Miyazaki University has been advancing DX education initiatives, and we incorporated Datachemical LAB into one of our laboratory exercises. Previously, there had been plans to have students write Python code, but the time and technical constraints made that impractical.

Yoshimaru:

And Datachemical LAB was able to bridge that gap.

Prof. Oshima:

Exactly. This time, we had students work through the full process — from model construction to validation — using "prediction of aqueous solubility of organic compounds" as the theme. For many students, it was a completely different experience from conventional lab work.In particular, being able to retrieve data instantly from a CSV file, and the workflow of extracting features from molecular structures to make predictions — all of that seemed fresh to them. You could see students genuinely surprised, thinking, "We can actually do this?"

Prediction of logS (water solubility) of organic compounds

Using the structures (SMILES notation) and water solubility data of 800 existing organic compounds as learning data, we performed machine learning predictions using the following steps.

① Descriptor calculation : 208 molecular descriptors are generated using the RDKit from the structures of organic compounds in the training data.

② Preprocessing : Select features, remove unnecessary descriptors, and organize the data into one suitable for analysis.

3) Data visualization : Understanding data features using heat maps of correlation coefficients and scatter plots

④ Model building and prediction : Build a predictive model using the training data and predict the logS of the compound each participant selected as their research subject.

⑤ Verification : Compare the predicted results with literature values and consider the relationship with the structure.

Experimental exercise using Datachemical LAB

Yoshimaru:

What do you see as the key challenges in applying machine learning to research and education?

Prof. Oshima:

Data volume is a real challenge. Experimental research is inherently limited in the number of data points available. Compared to research that generates tens of thousands of data points through high-throughput methods, the conditions are inevitably more restrictive.

The solvent screening work we do in metal extraction happened to be a theme where relatively large datasets could be collected, which made it well-suited to machine learning. So now, I'm always thinking about what kinds of research themes can realistically generate enough data and are amenable to modeling.

Yoshimaru:

It really comes down to having the perspective to assess, based on the nature of the research and its practical realities, how much data can actually be obtained. And it is precisely in those settings — where data is limited — that we believe Datachemical LAB can make a real contribution. We have continually refined our features to enable accurate predictions even with small datasets, and we look forward to continuing to expand the possibilities in this field through close collaboration with research and education communities.

Yoshimaru:

Finally, do you have any words for students who are beginning to study data science?

Prof. Oshima:

First and foremost: don't over-rely on AI or machine learning. Even in our student exercise, the prediction accuracy was good — but the model was far from perfect. And that's fine. In fact, experiencing firsthand that "things don't always work out" is precisely the point.

Yoshimaru:

I see — that's what it really means to master these tools.

Prof. Oshima:

I think so. Machine learning is an extension of statistics, so results aren't everything — what matters is how you think from there. As AI advances and the importance of data science grows, an exclusive reliance on conventional theoretical and scientific approaches will no longer suffice. Actively incorporating statistical methods is becoming indispensable. So I want students to carry a genuine awareness of "mastering AI." Used well, it is a powerful tool — but it should never be something that controls you. I want them to engage proactively, with a healthy sense of distance.
Looking ahead, I also hope to develop more practical educational package materials with a view toward broader use in academic settings.

Yoshimaru:

That sounds like a wonderful initiative. We at Datachemical would love to provide more opportunities for Datachemical LAB to be used broadly in university education — and we would be very glad to support your efforts in developing those materials.

Prof. Oshima:

I would greatly appreciate that.

Yoshimaru:

Thank you so much. Your sincere dedication to both research and education has been a great source of inspiration for all of us.

Prof. Oshima:

Thank you as well. It has been a pleasure.

Currently, we are offering paid academic users the opportunity to use Datachemical LAB for educational purposes.

Company Overview

Company Name: Datachemical Inc.

Address: Kuwano Building 2F, 6-23-4 Jingumae, Shibuya-ku, Tokyo 150-0001, Japan

Phone: +81-3-6778-2045 CEO: Shogo Yoshimaru

Site Menu

Datachemical LAB

Datachemical, Inc.

Datachemical LAB Case Study Interview

From the research and education fields

Professor, Chemistry and Life Science Program, Faculty of Engineering, University of Miyazaki

Data Chemical Co., Ltd. Representative Director and CEO

"No-code x Materials Informatics" spreading to research and education

Guest

Prof. Tatsuya Oshima — Professor, Chemical Life Program, Faculty of Engineering, University of Miyazaki

Interviewer

Shogo Yoshimaru — CEO, Datachemical Inc.

What prompted me to start using machine learning in my research?

Student-led machine learning expands through Datachemical LAB

The results seen in the field of education and the reality of data utilization

Experimental Science and Machine Learning: How to Deal with "Data Shortage"

How to deal with AI and machine learning: The attitude students need

Yoshimaru (Moderator):

Thank you for your time today. It has been about three years since your university adopted Datachemical LAB. To start, could you tell us how you have been using it in your own research?

Prof. Oshima:

Yoshimaru:

And that's when you turned to machine learning?

Prof. Oshima:

Yoshimaru:

How did things change after you introduced Datachemical LAB?

Prof. Oshima:

Yoshimaru:

Has this translated into tangible results?

Prof. Oshima:

We started with an extraction rate prediction model that achieved an R² of around 0.94. Since then, the volume of data students handle has grown and our validation methods have evolved. We are now building models with R² values of 0.98 to 0.99 — quite high accuracy.

Yoshimaru:

Your university has signed a partnership agreement with us with the aim of collaborating on education. As part of that, this year you actually used Datachemical LAB in the classroom.

Prof. Oshima:

Yoshimaru:

And Datachemical LAB was able to bridge that gap.

Prof. Oshima:

Prediction of logS (water solubility) of organic compounds

Using the structures (SMILES notation) and water solubility data of 800 existing organic compounds as learning data, we performed machine learning predictions using the following steps.

① Descriptor calculation : 208 molecular descriptors are generated using the RDKit from the structures of organic compounds in the training data.

② Preprocessing : Select features, remove unnecessary descriptors, and organize the data into one suitable for analysis.

3) Data visualization : Understanding data features using heat maps of correlation coefficients and scatter plots

④ Model building and prediction : Build a predictive model using the training data and predict the logS of the compound each participant selected as their research subject.

⑤ Verification : Compare the predicted results with literature values and consider the relationship with the structure.

Yoshimaru:

What do you see as the key challenges in applying machine learning to research and education?

Prof. Oshima:

Data volume is a real challenge. Experimental research is inherently limited in the number of data points available. Compared to research that generates tens of thousands of data points through high-throughput methods, the conditions are inevitably more restrictive.

Yoshimaru:

Yoshimaru:

Finally, do you have any words for students who are beginning to study data science?

Prof. Oshima:

First and foremost: don't over-rely on AI or machine learning. Even in our student exercise, the prediction accuracy was good — but the model was far from perfect. And that's fine. In fact, experiencing firsthand that "things don't always work out" is precisely the point.

Yoshimaru:

I see — that's what it really means to master these tools.

Prof. Oshima:

Yoshimaru:

That sounds like a wonderful initiative. We at Datachemical would love to provide more opportunities for Datachemical LAB to be used broadly in university education — and we would be very glad to support your efforts in developing those materials.

Prof. Oshima:

I would greatly appreciate that.

Yoshimaru:

Thank you so much. Your sincere dedication to both research and education has been a great source of inspiration for all of us.

Prof. Oshima:

Thank you as well. It has been a pleasure.

Currently, we are offering paid academic users the opportunity to use Datachemical LAB for educational purposes.

Company Overview

Datachemical LAB ​

​企業概要

Datachemical LAB ​

Prof. Tatsuya Oshima —
Professor, Chemical Life Program, Faculty of Engineering, University of Miyazaki

Shogo Yoshimaru —
CEO, Datachemical Inc.

Datachemical LAB

企業概要

Datachemical LAB