Advances in AI’s Future in Toxicology: Integrating Computational Prediction and Clinical Translation Through Explainable Artificial Intelligence

Artificial intelligence (AI) and machine learning are increasingly used in toxicological risk assessment to predict chemical toxicity, identify hazardous compounds, and support regulatory decision-making. However, the widespread adoption of these models is limited by their "black-box" nature, which reduces interpretability, transparency, and regulatory confidence. Explainable artificial intelligence (XAI) has emerged as a promising approach to address these challenges by revealing how input features, including chemical structures, exposure levels, and biological pathways, contribute to toxicity predictions. Techniques such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) provide interpretable insights into model decision-making and enhance trustworthiness, accountability, and mechanistic understanding. Despite these advances, most XAI applications in toxicology remain at the computational or preclinical stage, with limited clinical and regulatory translation. This review summarizes current AI-based approaches for toxicity prediction, examines the interpretability challenges that hinder their practical implementation, and highlights the emerging role of XAI in bridging the gap between computational prediction and clinical application. Greater standardization, integration of human data, and collaboration among academia, industry, and regulatory agencies will be essential for advancing transparent and clinically actionable toxicology models.