Current status and problems of in silico toxicity prediction based on computational toxicology

Abstract

Computational toxicology techniques can be used to construct ultra-high-throughput prediction systems capable of distinguishing toxic substances from a large number of candidate compounds. However, in the creation of high-accuracy in silico prediction models, problems are encountered, such as the low number of actual measured toxicity values. In this presentation, I will introduce a drug hepatotoxicity/pulmonary toxicity prediction system built to address these problems in the AMED project: “Construction of drug discovery informatics system.” In addition, I will introduce the AI-SHIPS project in the Ministry of Economy, Trade and Industry.