A novel feature extraction model for traffic injury severity and its application to Fatality Analysis Reporting...

Area of Science:

• Transportation Safety
• Traffic Management
• Injury Prevention

Background:

• Severe injuries in traffic crashes remain a critical issue in transportation safety.
• Identifying factors influencing crash severity is essential for effective injury prevention strategies.

Purpose of the Study:

• To propose an improved Markov Blanket algorithm for extracting significant factors affecting traffic crash injury severity.
• To evaluate the performance of the proposed algorithm in identifying key injury predictors.
• To analyze the relationship between specific factors and injury severity.

Main Methods:

• Utilized Fatality Analysis Reporting System (FARS) data for analysis.
• Developed an improved Markov Blanket algorithm to identify significant factors from 29 variables.
• Applied Pearson correlation coefficient test and classification algorithms (Bayesian networks, C4.5 decision tree) for validation.

Main Results:

• The improved Markov Blanket algorithm successfully extracted significant impact factors influencing crash injury severity.
• The algorithm demonstrated improved classification accuracy and reduced classification error rates.
• Identified key factors associated with severe injuries: bad weather, nighttime crashes, drunk driving, single-driver incidents, and distracted driving.

Conclusions:

• The improved Markov Blanket algorithm is effective in identifying critical factors for traffic injury severity.
• Factors such as adverse weather, nighttime, impaired driving, and driver distraction significantly increase the risk of severe injuries.
• Findings can inform targeted interventions to enhance road safety and reduce severe crash outcomes.