Weighted network analysis of adverse outcome pathways decodes the multiscale mechanisms of environmental toxicity.

The rapid proliferation of synthetic chemicals has significantly outpaced traditional toxicity characterization, leaving a critical data gap in environmental health risk assessment. While the adverse outcome pathway (AOP) framework provides a mechanistic scaffold for organizing toxicity knowledge, it is currently limited by a focus on linear pathways and a bias toward well-studied endpoints. Conversely, the exposome paradigm captures broad environmental stressors but often lacks the mechanistic depth required for causal interpretation. A fundamental challenge remains in developing integrative paradigms that can systematically bridge these multi-scale datasets to decode complex, chemical-induced diseases. Here we show that AOP-ExpoVis, an integrative computational platform, synergizes exposome-disease networks with AOP ontologies to prioritize pathogenic mechanisms through a weighted phenotype-disease scoring algorithm. By integrating chemical, gene, phenotype, and disease associations, the platform identifies key phenotypes and maps them to curated pathways to generate testable mechanistic hypotheses. Validation across three distinct case studies involving 2,2',4,4'-tetrabromodiphenyl ether (BDE-47), arsenic, and perfluoroalkyl substances (PFAS) demonstrated that AOP-ExpoVis accurately identifies both conserved and chemical-specific toxic pathways, such as aryl hydrocarbon receptor activation and lipid metabolism disruption. AOP-ExpoVis provides an open-source tool for rapid mechanistic inference that overcomes the limitations of traditional, single-pathway frameworks. This work advances predictive toxicology by enabling the systematic prioritization of chemical hazards and the refinement of regulatory risk assessment in a data-rich environment.