Large language models for spreading dynamics in complex systems

2026-02-08

Physics reports

AI Summary

This work proposes a novel paradigm for modeling information and disease propagation by integrating large language models (LLMs) as endogenous intelligent agents within complex network and epidemiological frameworks. Traditional diffusion dynamics struggle to incorporate multidimensional semantic factors such as message framing, cultural context, and cognitive biases. By leveraging the natural language understanding, reasoning, and generation capabilities of LLMs, this approach enables semantic-aware modeling, monitoring, and prediction of both digital epidemics (e.g., misinformation) and biological epidemics (e.g., infectious diseases). The method overcomes the limitations of conventional models in processing unstructured semantic content, establishing a new framework that fuses semantic perception with dynamic feedback. This significantly enhances modeling accuracy, early detection sensitivity, and intervention efficacy, while also outlining a clear interdisciplinary research trajectory and identifying key challenges.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Data Mining & Knowledge Management: Semantic Web

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Problem

• spreading dynamics
• complex systems
• large language models
• epidemic modeling
• information propagation

Innovation

• large language models
• spreading dynamics
• complex systems
• epidemic modeling
• contextual reasoning