A research team from the Institute for Materials Chemistry and Engineering, Kyushu University, developed a method to automatically construct deep learning models that predict transition states of chemical reactions. They demonstrated that transition states can be appropriately predicted even in systems with many atoms, and consistent results are obtained across various deep learning models.

Key achievements:

• Automated construction of deep learning models
• Efficient prediction of transition states in reactions with multiple solvent molecules
• Potential application of artificial intelligence to elucidate complex chemical reactions such as enzyme reactions