Host: The Japanese Society of Toxicology
Name : The 47th Annual Meeting of the Japanese Society of Toxicology
Date : 2020 -
Existing in silico software often fails to predict mutagenicity of our in-house compounds. Therefore, we have developed an AI based prediction system for mutagenicity by using both chemical structure images and Ames test results as a training dataset. We were able to improve the prediction accuracy of in-house compounds with the AI system compared to that of existing in silico software. Furthermore, the AI prediction system can visualize the rationale for the prediction, so it can also be utilized in ICH-M7 expert review. We would like to introduce how this AI-based prediction system works.