Evaluating the Safety Risk of Rural Roadsides Using a Bayesian Network Method.

Area of Science:

• Road safety engineering
• Transportation risk assessment
• Geographic information systems

Background:

• Evaluating rural roadside safety is crucial for efficient budget allocation and avoiding unnecessary safety installations.
• Limited or missing crash data often hinders accurate safety risk assessment in rural areas.

Purpose of the Study:

• To propose and validate a Bayesian Network (BN) method for assessing rural roadside safety risks using expert judgment.
• To evaluate the effectiveness of incorporating 'access point density' as a novel safety risk factor.

Main Methods:

• Developed a Bayesian Network (BN) model incorporating expert judgments on conditional probabilities of safety risk factors.
• Included seven established factors and a new factor, access point density.
• Validated the model using a case study of rural road networks in Nantong, China, comparing results with historical run-off-road (ROR) crash data.

Main Results:

• The BN method successfully identified road segments with higher safety risks, which correlated significantly with higher crash severity.
• The inclusion of access point density notably improved the accuracy of safety risk evaluation.
• The proposed method demonstrated a statistically significant correlation between identified high-risk segments and actual crash severity.

Conclusions:

• The Bayesian Network (BN) method offers a low-cost, accurate solution for evaluating rural roadside safety, particularly where crash data is incomplete.
• Access point density is a significant contributing factor to rural roadside safety risks.
• This approach is highly valuable for regions with extensive rural road networks and data limitations.