Molecular fingerprint and prediction of pharmacological activity using skeletal structural formula images

A new form of drug discovery research that makes use of artificial intelligence.To improve AI prediction performance, we added a new method of expression that provides a bird's-eye view of the entire structure of a compound, in addition to the conventional way of expressing compounds. The combination of these various methods of representation makes it possible to improve the predictive performance of artificial intelligence.

I thought it is new that the activity prediction in drug development, which used to be done manually by intuition and experience, is now done by machine learning, and that efficiency and accuracy have been improved by using two types of images: molecular fingerprints and skeletal structure formula images.
The paper also made me think that in the future, people will be able to participate in drug development through apps, etc., even if they do not know much about drugs.

Ayuko Takagi(Sony Institute of Higher Education Shohoku College Dept. of Business & Informatics Associate Professor)

I believe that the research is socially significant and highly reliable using scientific methods.
The research results have been presented at domestic and international conferences with advisors outside of the university, and academic contributions are also expected.
It would have been better if it had presented a unique perspective of Digital Hollywood.

Yoshiko Goda(Kumamoto Univ. Research Center for Instructional Systems Associate Professor)