Real-time crash prediction models: State-of-the-art, design pathways and ubiquitous requirements

Area of Science:

• Traffic Safety Engineering
• Artificial Intelligence in Transportation
• Data Science for Mobility

Background:

• Proactive traffic safety management systems aim to prevent accidents by monitoring real-time conditions and intervening.
• Advancements in AI and sensor technology are enabling the development of these systems.
• Reliable real-time crash prediction models are essential prerequisites for proactive safety management.

Purpose of the Study:

• To conduct a comprehensive review of existing real-time crash prediction models.
• To identify state-of-the-art approaches, synthesize findings, and address challenges in model development.
• To facilitate the translation of crash prediction models into operational proactive traffic safety systems.

Main Methods:

• Systematic literature review incorporating text mining and topic modeling techniques.
• Analysis of development pathways and design requirements from existing studies.
• Evaluation of the universality and design compatibility of current crash prediction models.

Main Results:

• Identified discrete and fragmented development efforts due to varying modeling approaches.
• Synthesized ubiquitous design requirements for real-time crash prediction models.
• Assessed the compatibility and universality of existing models for practical application.

Conclusions:

• Significant challenges remain in formulating a ubiquitous proactive safety management system.
• Existing models require further development and standardization to be implemented in real-world scenarios.
• This review serves as a knowledge resource to accelerate the transition from concept to operational safety technology.