“AI Prince Shōtoku” classify multiple voices and directs material development

June 23, 2025

Accelerating the design of environmentally friendly plastics with diverse property requirements

A research team at the RIKEN CSRS has developed a method that uses artificial intelligence (AI) to extract multiple types of information from a single measurement obtained via time-domain nuclear magnetic resonance (TD-NMR). Nicknamed the “AI Prince Shōtoku” for its ability to simultaneously interpret multiple signals and provide guidance for material development, this system is expected to accelerate the material design cycle for environmentally friendly plastics.

Biodegradable plastics are key materials for achieving a circular economy through resource recycling. However, the polymer chains that form plastics exhibit diverse tertiary structures, and differences in their higher-order structure and dynamics greatly affect physical properties. As a result, the high cost of analysis and the time-consuming process—taking over 30 days to evaluate multiple material properties—have been major bottlenecks in development. To address this, the research team established a new method that applies convolutional neural networks (CNNs), a type of AI, to TD-NMR data, which can be obtained in under 30 minutes. The approach denoises the data and extracts molecular dynamics features related to material toughness and flexibility. This “AI Prince Shōtoku” is expected to contribute to streamlining and reducing the cost of material development processes.

 

Original article

npj Materials Degradation　doi: 10.1038/s41529-025-00613-7

R. Fujita, Y. Amamoto, J. Kikuchi,

"Bayesian Optimization of Biodegradable Polymers via Machine Learning Driven Features from Low-Field NMR Data".

Contact

Jun Kikuchi
Team Director
Environmental Metabolic Analysis Research Team