Showa Denko Establishes Integrated Data Pipeline to Collect, Format, and Accumulate Experimental Data and Analyze Them with AI
- SDK has established data pipeline utilizing electronic lab notebooks. In less than a year of operation, there have already been some cases in which the data pipeline has shortened the product development period by two months -
November 02, 2022
Showa Denko K.K.
Showa Denko K.K. (SDK) (TOKYO: 4004) (President and CEO: Hidehito Takahashi) has established a data pipeline that integrates electronic lab notebooks into which materials scientists input raw experimental data, a database management system, and a materials informatics web application (MI app) ahead of the company’s competitors. The MI app incorporated into the data pipeline allows data scientists to have easy access to raw experimental data, which is used to build artificial-intelligence-based models (AI models).
By using the electronic lab notebooks, the data pipeline enables data scientists to have access to qualitative information, which is previously lost in the data formatting process for building the AI models. Suppose a data scientist, who manages an MI app, has doubts about AI predictions. In that case, the data scientist can quickly review raw experimental data and find factors that misled the AI model, including abnormal data. In addition, information recorded on electronic lab notebooks is automatically extracted and is used for building the AI models by utilizing MLOps (Machine Learning Operations)*1, which SDK established before. Thus, materials scientists can always use a high-accuracy MI app that reflects the latest experimental data.
By utilizing the data pipeline, which includes the MI app, SDK will accelerate materials development and continue providing high-value-added products that meet the market demand.
■ SDK has established its global presence, being listed among the 30 “Materials Informatics’ Key Players,” hosted by Lux Research Inc.
SDK has been developing its original MI app. It can predict compound properties under a particular formulation of raw materials or propose candidate formulations that satisfy required properties with AI models. The AI models are built based on past experimental data. SDK has been utilizing the MI app to promote research and development activities in almost all business divisions on a daily basis. As a result, SDK has established its global presence, being listed among the 30 entities under “Materials Informatics’ Key Players,” hosted by Lux Research Inc.*2
■ With a click on the link, data scientists can quickly access raw experimental data
In December 2021, SDK established a data pipeline that collects, formats, and accumulates experimental data and analyzes the data with AI models by integrating electronic lab notebooks named “BIOVIA Notebook>*3,” a database management system, and an MI app.
The data pipeline allows data scientists to quickly access raw experimental data stored in electronic lab notebooks by clicking the link displayed on the MI app. As a result, data scientists, who manage the MI app, can build AI models more accurately by confirming qualitative information in raw experimental data. The raw experimental data include spectral data that shows materials properties or molecular structure images, previously lost in the data formatting process. By judging the validity of raw experimental data, the data scientists can select appropriate data for improving the accuracy of the AI models.
■ The data pipeline enables the MI app to reflect the latest experimental data automatically
In addition, the data pipeline has functions to extract raw experimental data stored on electronic lab notebooks and format the data to build AI models by utilizing MLOps. Thus, by just uploading information onto electronic lab notebooks, materials scientists can use a high-accuracy MI app that always reflects the latest experimental data without manually formatting the data or tuning the AI models.
■ In some cases, the data pipeline has shortened the product development period by two months
Although we started the operation of the new data pipeline just a little less than a year ago, there have already been some cases in which the data pipeline has shortened the product development period by two months. For example, when we develop an adhesive for electronic devices which is required to have strong adhesion and easiness in detachment simultaneously, our materials scientists aim to develop an adhesive that has both adhesive strength and no adhesive residue, though these properties have a trade-off relationship. By utilizing the MI app of the data pipeline and focusing development efforts on the promising formulation of raw materials, our materials scientists successfully reduced the number of times of experiments to one-third of those with conventional methods, thereby shortening the time for experiments by two months, namely, from three months to one month.
■ The data pipeline supports the spread of electronic lab notebooks
Development of the data pipeline has been promoting the spread of electronic lab notebooks among the Showa Denko Group’s materials scientists. As of September 2022, the number of registered IDs for electronic lab notebooks was about 400. We expect this number will increase to about 700 by the end of 2023.
● Comment made by SDK’s staff responsible for the spread of electronic lab notebooks
“We think experimental data are an important technical asset of the company. Therefore, accumulating information in electronic lab notebooks is effective for us in the accumulation, sharing, succession, and utilization of information. We often hear that other chemical companies also have difficulty spreading electronic lab notebooks. In the SDK Group, we also heard many opinions saying, “It takes time,” or “we don’t find any advantage of using electronic lab notebooks.” Therefore, we had a fact-finding survey based on interviews with each department. Based on the survey, we added originally developed functions to electronic lab notebooks and solved many problems through this effort, thereby promoting the spread of electronic lab notebooks.”
● Comment made by Dassault Systèmes, the provider of an electronic lab notebook system named “BIOVIA Notebook”
“Establishment of the integrated data pipeline, including our electronic lab notebook system, is an advanced example of the realization of the data-driven R&D process in Japan. We will continue contributing to the promotion of this MI project by providing our electronic lab notebook solution for Showa Denko, which has the leading number of registered IDs for electronic lab notebooks in the chemical industry of Japan.”
■ SDK will have an online presentation about the data pipeline
SDK will have an online session to present the data pipeline in an online event, “BIOVIA USER GROUP MEETING JAPAN 2022,” hosted by Dassault Systèmes.
- Schedule of our presentation: 4:05 p.m. – 4:30 p.m. JST, Wednesday, November 9, 2022
- Presentation title: Utilization of electronic lab notebooks at Showa Denko
Note: To watch the live-streaming of the presentation, please register in advance
This presentation will be held in Japanese.
SDK will continue committing itself to utilize computational science, MI, and AI, accelerating the materials development and pioneering in creating functions that meet the needs of the times, thereby contributing to society.
- ※1 For detail, please refer to SDK’s news release, “Showa Denko Introduces Machine Learning Operations into AI-based Prediction Systems for Accelerating Materials Development,” announced on April 21, 2022.
- ※2 “Materials Informatics’ Key Players” by Lux Research Inc.https://www.luxresearchinc.com/blog/materials-informatics-key-players/
- ※3 For detail on BIOVIA Notebook, please refer to
BIOVIA is commercial trademark or registered trademark of Dassault Systèmes, a French “société européenne” (Versailles Commercial Register # B 322 306 440), or its subsidiaries in the United States and/or other countries.
For further information, contact:
Public Relations Group, Brand Communication Department (Phone: 81-3-5470-3235)
For further information, contact
Public Relations Group, Brand Communication Department