AI-Driven Framework Predicts Traffic Crash Severity for Improved Road Safety

Strategic Insights -

Researchers at Jouf University in Sakaka, Saudi Arabia, have developed an AI-driven machine learning (ML) framework to predict traffic crash severity. By analyzing a large-scale dataset of over 2.26 million records, the framework integrates human, crash-specific, and vehicle-related factors to enhance predictive accuracy and reliability. The proposed framework, which incorporates feature engineering and clustering techniques, demonstrated superior performance, achieving 96.19% accuracy and an F1-score (macro) of 95.28%.

Key Takeaways:

  • The AI-driven framework utilizes a large-scale dataset of over 2.26 million records to predict traffic crash severity.
  • The framework integrates human, crash-specific, and vehicle-related factors to enhance predictive accuracy and reliability.
  • The methodology incorporates feature engineering, clustering techniques, and oversampling methods to address class imbalance.
  • The Extra Trees (ET Classifier) ensemble model demonstrated superior performance, achieving 96.19% accuracy and an F1-score (macro) of 95.28%.
  • The framework provides a scalable, AI-powered solution for traffic safety, offering actionable insights for intelligent transportation systems (ITS) and accident prevention strategies.
  • The research aims to enhance traffic risk assessment and data-driven decision-making.
  • The study involves researchers from Jouf University, including Ayman Mohamed Mostafa, Bader Aldughayfiq, Mayada Tarek, Alaa S. Alaerjan, Hisham Allahem, Murtada K. Elbashir, Mohamed Ezz, and Eslam Hamouda.
  • The research was published in Scientific Reports, a journal published by Nature Portfolio.

Statistics:

  • 2.26 million records analyzed in the dataset.
  • 96.19% accuracy achieved by the Extra Trees (ET Classifier) ensemble model.
  • 95.28% F1-score (macro) achieved by the ET Classifier.
  • The research aims to reduce risks and enhance mobility through improved road safety.

Sources:

  • AI-based prediction of traffic crash severity for improving road safety and transportation efficiency. Scientific Reports, 2025;15(1):27468.
  • Nature Publishing Group - www.nature.com/.
  • Scientific Reports - www.nature.com/srep/.
  • Jouf University - 72388, Sakaka, Saudi Arabia.