Artificial Intelligence for Drug Toxicity and Safety

This chapter reviews the use of machine learning and artificial intelligence algorithms to evaluate drug safety from discovery to post-launch. Machine learning techniques have been a part of safety evaluations since quantitative structure activity relationship (QSAR) modeling but are becoming leveraged at every stage of development to improve decision-making using data-driven approaches. This has been spurred on by the evolving approaches used to evaluate drugs in development and the rapid advancement of novel algorithms to learn from prior data. This growth does not come without challenges, particularly around supplying these algorithms with the prior knowledge to make accurate determinations. This review will highlight the algorithms being used, new use cases, where they have met challenges and successes, and look toward the future of Artificial Intelligence (AI) in evaluating drug toxicity and safety.