A <sc>bstract</sc>
The Adverse Outcome Pathway (AOP) framework is a cornerstone of 21st-century toxicology, providing a structured method for organizing mechanistic knowledge to support risk assessment. However, the inherent complexity of biological systems, characterized by interconnected signaling networks, reveals the limitations of simple, linear AOP representations. This has created a pressing need for advanced computational tools that can visualize and analyze these intricate AOP networks. This study introduces the AOP Network Viewer, a novel software platform designed to address the critical challenges of scalability, interactivity, and deep analysis in toxicological research. The system is built on a scalable client-server architecture with a graph database backend, supporting sophisticated interactive features such as multi-pathway selection, dynamic path-finding, and advanced filtering. The principal innovation of the AOP Network Viewer is its integrated AI Analysis Panel, a Large Language Model (LLM)-powered engine that provides a deep, context-aware analysis of selected network components. This AI system utilizes a unique three-layer prompt architecture to deliver domain-specific insights across seven specialized analytical modes. Case studies demonstrate the platform’s utility in developing alternative testing strategies for endocrine disruptors, interpreting high-throughput screening data, and generating hypotheses for toxicity assessment of mixtures. The AOP Network Viewer is engineered to translate complex user queries into focused, interpretable biological networks, transforming AOP network visualization from a passive viewing experience into an interactive and intelligent analytical process, thereby accelerating mechanistic discovery and enhancing evidence-based regulatory decision-making. The app is source code is available at https://github.com/abhik1368/aop-visualizer-clean .