Acceptance of in silico models in regulatory toxicology

Computational toxicology is an increasingly important field for toxicological assessment, as traditional in vivo testing faces ethical, financial, and practical limitations. Quantitative structure–activity relationship models, read-across approaches, mechanistic models, and machine-learning methods can help predict toxicity and reduce reliance on animal testing. Yet strong scientific performance alone does not ensure regulatory acceptance. Questions remain about how models should be validated, how prediction reliability and uncertainty should be assessed and reported, and how computational evidence should be compared with laboratory and animal data. Recent improvements in datasets, methods, and the use of biological and mechanistic information have strengthened the capabilities of many models. However, differences in data quality, reporting practices, validation expectations, and regulatory requirements continue to challenge their broader use. Clearer guidance and closer collaboration among researchers, toxicologists, and regulators are needed to address these gaps and build confidence in computational evidence.This Research Topic aims to examine how in silico models can be validated, interpreted, and integrated into regulatory toxicology. It will explore the scientific, methodological, and policy requirements for increasing confidence in computational predictions and supporting their use in toxicological assessment and regulatory decision-making. Contributions may investigate approaches to characterizing uncertainty, combining computational evidence with experimental and mechanistic information, and accommodating rapidly developing technologies within regulatory frameworks. The Research Topic also seeks to identify barriers to implementation and highlight examples of regulatory submissions or decisions supported by in silico data.To gather further insights into the validation, interpretation, integration, and regulatory implementation of in silico models, we welcome contributions on toxicological assessment and regulatory decision-making, including, but not limited to, the following themes:• Validation frameworks for computational toxicology models• Approaches to validating machine-learning and artificial intelligence models• Quantification, communication, and management of model uncertainty• Applicability domains, data quality, and chemical-space coverage• Explainability, transparency, reproducibility, and standardized reporting• Quantitative structure–activity relationship models and read-across approaches• Integration of in silico evidence into integrated approaches to testing and assessment• Use of adverse outcome pathways and mechanistic information in computational models• Case studies of in silico data submitted to or accepted by regulatory authorities• Comparative analyses of international regulatory frameworks and policies• Strategies for building confidence and stakeholder acceptance• Challenges in applying computational methods to emerging substances, products, and materialsTo promote reproducibility, transparency, and trust, authors are strongly encouraged to align submissions with FAIR principles by making research data, models, and code Findable, Accessible, Interoperable, and Reusable. Submissions of datasets and models as FAIR² Data Articles are particularly welcome. Unlike conventional articles focused on hypothesis-driven analysis, FAIR² Data Articles focus on high-value datasets—their structure, provenance, curation, and potential for reuse. These peer-reviewed, citable articles can provide academic recognition for dataset creators, make thoroughly documented and machine-readable resources available for regulatory scrutiny and benchmarking, and support community reuse through clear provenance and standardized metadata. This can help curated toxicological datasets serve as resources for training, testing, and validating in silico tools. Topic editors: Antreas Afantitis, Todd Zurlinden, Iseult Lynch Source: https://www.frontiersin.org/research-topics/85771/acceptance-of-in-silico-models-in-regulatory-toxicology