CrashSage: A Large Language Model-Centered Framework for Contextual and Interpretable Traffic Crash Analysis

2025-05-08

AI Summary

With over 1.3 million traffic crash fatalities annually worldwide, there is an urgent need for interpretable, data-driven crash analysis to inform effective road safety interventions. Method: This work introduces the first large language model (LLM)-based interpretable crash analysis framework. It overcomes the limitation of traditional statistical models in leveraging unstructured crash narratives by innovatively transforming tabular crash data into semantically rich textual representations. The framework incorporates fidelity-preserving contextual enhancement and integrates gradient-based explainability methods (e.g., Integrated Gradients) to enable fine-grained severity prediction and multi-level attribution analysis. Contribution/Results: Fine-tuned on LLaMA3-8B, the framework significantly outperforms strong baselines—including GPT-4o and LLaMA3-70B—in crash severity prediction. It supports both instance-level causal attribution and macro-level risk factor identification, thereby enhancing the precision and operationality of road safety policy and intervention design.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Computer Vision: Large Vision Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Road crashes claim over 1.3 million lives annually worldwide and incur global economic losses exceeding $1.8 trillion. Such profound societal and financial impacts underscore the urgent need for road safety research that uncovers crash mechanisms and delivers actionable insights. Conventional statistical models and tree ensemble approaches typically rely on structured crash data, overlooking contextual nuances and struggling to capture complex relationships and underlying semantics. Moreover, these approaches tend to incur significant information loss, particularly in narrative elements related to multi-vehicle interactions, crash progression, and rare event characteristics. This study presents CrashSage, a novel Large Language Model (LLM)-centered framework designed to advance crash analysis and modeling through four key innovations. First, we introduce a tabular-to-text transformation strategy paired with relational data integration schema, enabling the conversion of raw, heterogeneous crash data into enriched, structured textual narratives that retain essential structural and relational context. Second, we apply context-aware data augmentation using a base LLM model to improve narrative coherence while preserving factual integrity. Third, we fine-tune the LLaMA3-8B model for crash severity inference, demonstrating superior performance over baseline approaches, including zero-shot, zero-shot with chain-of-thought prompting, and few-shot learning, with multiple models (GPT-4o, GPT-4o-mini, LLaMA3-70B). Finally, we employ a gradient-based explainability technique to elucidate model decisions at both the individual crash level and across broader risk factor dimensions. This interpretability mechanism enhances transparency and enables targeted road safety interventions by providing deeper insights into the most influential factors.

Problem

• Analyzing road crashes with contextual and interpretable insights
• Overcoming limitations of traditional crash data analysis methods
• Enhancing crash severity prediction using LLM-based techniques

Innovation

• Tabular-to-text transformation for enriched crash narratives
• Context-aware data augmentation using base LLM model
• Fine-tuned LLaMA3-8B model for crash severity inference