The AI Trends in Chemical Space for Drug Discovery

Pharmaceutical development involves a substantial amount of time and expense. In recent years, drug discovery using AI technology is expected to accelerate pharmaceutical development. The primary goal of drug discovery is to expedite the discovery of new drugs at a lower cost by constructing AI models using chemical structures and experimental data obtained from compound-related databases. Nevertheless, there are limitations to the available experimental data that can be acquired as training data, and their comprehensiveness is limited. The development and anticipated effectiveness of drug discovery foundation models are expected to address this challenge. These models effectively learn compound data through pre-training techniques, such as self-supervised learning, similar to those used in recent large language models and image AI. This chapter provides an overview of cutting-edge AI research in medicinal chemistry, elucidating the current status and efforts to overcome the challenges in this field.