Artificial Intelligence in Drug Toxicity Prediction: Recent Advances, Challenges, and Future Perspectives

Area of Science:

• Pharmacology and Toxicology
• Computational Chemistry
• Drug Discovery and Development

Background:

• Drug toxicity is a major cause of late-stage failures, with over 30% of candidates discarded.
• Accurate toxicity prediction is crucial for identifying safe and effective human therapeutics.
• Traditional methods are time-consuming and costly, necessitating advanced predictive approaches.

Purpose of the Study:

• To provide a comprehensive overview of recent advancements in AI-based drug toxicity prediction.
• To highlight machine learning and deep learning applications in predicting compound toxicity.
• To offer resources and discuss challenges for improving AI toxicity prediction models.

Main Methods:

• Review of machine learning algorithms and deep learning architectures for toxicity prediction.
• Analysis of AI applications for six major toxicity properties and Tox21 assay endpoints.
• Compilation of public data sources and toxicity prediction tools.

Main Results:

• AI offers more accurate and efficient methods for predicting potential toxic effects of drug candidates.
• Recent advances demonstrate the utility of various AI models in assessing compound safety.
• Identification of key challenges and future directions for AI-driven toxicity assessment.

Conclusions:

• AI-based toxicity prediction is vital for accelerating drug discovery and reducing attrition rates.
• Continued research and development in AI are essential for enhancing model performance and reliability.
• This review serves as a guide for researchers exploring AI applications in drug safety evaluation.