Computational Modeling for the Prediction of Hepatotoxicity Caused by Drugs and Chemicals

Hepatotoxicity presents a significant challenge for drug development and regulatory science. Despite many efforts to eliminate drugs as hepatotoxic before they are tested in humans, hepatotoxic drugs often escape preclinical toxicity testing and are not identified as such until they are in a later stage of drug development, sometimes not until after approval. The development of robust predictive models for evaluating the potential of liver injury in humans caused by drug candidates and chemicals is urgently needed. With the advance of in silico methods, high-throughput assays, and toxicogenomics, huge amounts of toxicity data have been generated, offering opportunities to utilize machine learning algorithms to build models for predicting hepatotoxicity. In this chapter, we will specifically introduce machine learning technologies, together with recent case studies. One of these case studies exemplifies the use of machine learning for resolving practical hepatotoxicity questions.