Expanding toxicology predictions: The evolution of QSAR and novel computational approaches

Quantitative structure–activity relationship (QSAR) modeling has long stood at the forefront of computational toxicology, enabling the prediction of biological activities and toxicities of chemicals based on their molecular structures. In an era marked by increasingly stringent regulatory requirements, high-throughput screening technologies, and the proliferation of chemical compounds, the demand for reliable, efficient, and interpretable computational toxicology models has never been greater. Despite significant advances, conventional QSAR approaches continue to face key limitations, particularly regarding data quality, model transferability, and the capacity to account for complex biological interactions. Recent research has witnessed the rise of novel machine learning techniques, deep learning, and integrative modeling strategies—heralding a paradigm shift toward more accurate and holistic toxicological prediction. Yet, challenges remain around model interpretability, validation, uncertainty quantification, and alignment with real-world biological and regulatory scenarios.

This Research Topic aims to critically explore the emerging frontiers in predictive toxicology by focusing on the evolution of QSAR and the integration of advanced computational models. It seeks to elucidate how novel algorithms, data curation strategies, and systems-level approaches can improve the efficacy, reliability, and applicability of toxicology predictions. The objective is to foster discussion around methodological advancements, highlight best practices for model development and validation, and assess the translational potential of computational predictions in both academic and regulatory toxicology.

To gather further insights in the fast-evolving landscape of computational toxicology, we invite articles that concentrate on, but are not limited to, the following themes:

Advances and challenges in traditional and novel QSAR modeling

Machine learning and deep learning applications in toxicology prediction

Integrative multi-omics and systems toxicology approaches

Strategies for data quality improvement, curation, and harmonization

Model interpretability and explainable artificial intelligence in toxicological contexts

Regulatory perspectives and practical applications of computational toxicology predictions

Case studies exemplifying the translation of computational models to experimental or regulatory settings

We welcome the submission of original research articles, reviews, brief reports, methodologies, perspectives, and case studies that address theoretical, methodological, or applied aspects at the intersection of computational modeling and toxicology prediction.