Unraveling the Causes of Fatal Crashes in the U.S.: A Machine Learning Approach to Safer Roads

This project investigates the underlying causes of fatal traffic crashes in the United States using advanced machine learning (ML) techniques to enhance road safety. Each year, traffic crashes claim over 42,000 lives nationwide, inflicting significant social, economic, and health burdens. Traditional analytical methods have struggled to capture the complex, nonlinear interactions among factors such as driver behavior, vehicle characteristics, roadway design, and environmental conditions. To address this limitation, this project employs data-driven ML models to identify key determinants of fatal crashes and generate actionable insights for evidence-based safety interventions. The research activities will proceed in four phases. First, comprehensive crash data will be collected from the National Highway Traffic Safety Administration (NHTSA) and integrated across multiple datasets to ensure completeness and consistency. Next, statistical analysis and visualization will be used to identify spatial and temporal trends in crash patterns, revealing geographic disparities and risk concentrations. In the modeling phase, several machine learning algorithms—Balanced Bagging, Balanced Random Forest, and RUSBoost—will be developed and compared against traditional logistic regression models to enhance prediction accuracy in imbalanced datasets. Finally, the top-performing model will be used to assess variable importance and generate policy-relevant recommendations. OBJECTIVE: The objective of this project is to develop predictive models that accurately identify risk factors associated with fatal crashes and support data-informed decision-making by transportation agencies. The findings will guide targeted interventions such as improved traffic regulations, safer roadway designs, and enhanced vehicle technologies. This research will provide a scalable analytical framework for improving transportation safety and sustainability nationwide.

• Record URL:
  • https://transportation.umkc.edu/project/unraveling-the-causes-of-fatal-crashes-in-the-us-a-machine-learning-approach-to-safer-roads

Language

• English

Project

• Status: Active
• Funding: $117,500.00
• Contract Numbers:

  69A3552348335

• Sponsor Organizations:

  Department of Transportation

  Office of the Secretary
  1200 New Jersey Avenue, SE
  Washington, DC  United States  20590
• Managing Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Performing Organizations:

  Center for Healthy and Durable Transportation

  University of Missouri Kansas City
  Kansas City, Missouri  United States  64110
• Principal Investigators:

  Wang, Yangmei

  Wang, Tiankai

• Start Date: 20260101
• Expected Completion Date: 20261231
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers Program

Subject/Index Terms

• TRT Terms: Crash analysis; Crash causes; Crash data; Crash risk forecasting; Fatalities; Logistic regression analysis; Machine learning; Predictive models; Statistical analysis
• Subject Areas: Data and Information Technology; Highways; Safety and Human Factors;

Filing Info

• Accession Number: 01986490
• Record Type: Research project
• Source Agency: Center for Healthy and Durable Transportation
• Contract Numbers: 69A3552348335
• Files: UTC, RIP
• Created Date: Apr 21 2026 1:45PM