Advancing car crash analysis

In traffic safety research, studies typically focus on factors such as road conditions, driver behavior, and vehicle type, as well as the resulting crash outcomes, like severity and frequency. However, the relationships between these factors may also play a crucial role in predicting crashes. This was the focus of Mahsa Jafari, a PhD student in Civil Engineering, whose presentation won top honors at the recent Transportation Research Board (TRB) Annual Meeting in Washington, D.C. Mahsa's research analyzes traffic crash data using a hybrid method that combines statistical techniques and artificial neural networks.

Her presentation, titled Investigating the Influence of Socioeconomic Factors on the Relationships Between Road Characteristics and Traffic Crash Frequency and Severity, offers a unique approach to understanding traffic safety. Using advanced statistical and machine learning methods, she is able to identify key variables affecting crashes and highlight the benefits of combining statistical models with artificial neural networks. Jafari’s unique approach has enabled her to identify that socioeconomic factors, such as neighborhood age demographics, income, and vehicle ownership, influence the frequency and severity of traffic crashes.

"The Structural Equation Modeling (SEM) analysis showed that factors like road conditions and vehicle ownership affected crash severity. However, most of the other factors didn’t have a significant impact," Jafari explains. "But when I used Artificial Neural Networks (ANN), they were able to uncover hidden patterns and showed that the interactions between different factors were very important in predicting crashes. This suggests that using ANN alongside SEM helps uncover insights that traditional methods might miss."

"This hybrid method helps us see both simple and more complex relationships between factors, which is important because traditional models can overlook them," she adds. The hybrid model gave a deeper understanding of crash causes, helping to identify more factors that contribute to crashes and providing better information for decision-making.

Her innovative approach is one of the first in the field to use this hybrid method in traffic safety research, which Jafari believes can offer valuable insights for future transportation policies and safety initiatives. "By focusing on neighborhood characteristics, we can design targeted educational programs or safety interventions based on the specific needs of different communities," she shares.

Jafari was introduced to TMU’s Civil Engineering program through a recommendation from a friend who had worked with her current supervisor, Dr. Bhagwant Persaud, a recognized expert in highway safety. She was drawn to Dr. Persaud’s work on the Highway Safety Manual and his extensive publications in the field. "I was eager to work with Dr. Persaud because of his expertise and the research opportunities at TMU," she notes.

Her presentation at TRB was well-received by professionals and peers in the traffic safety field. "I felt confident presenting because my research is grounded in addressing a gap in the literature, and I’ve been working on it for over two years," she shared. "Many studies focus on a set of independent and dependent variables, but they often don’t explore the relationships between them. That’s what I aimed to address."

Looking ahead, Jafari plans to continue expanding her research, with several papers currently under review. "I hope to encourage others in the field to adopt hybrid methods in their analysis, as relying on just one method can sometimes overlook important relationships."

Jafari’s research journey also highlights the value of interdisciplinary learning. She frequently draws insights from outside the transportation field, including psychology, to enhance her understanding of factors like driver behavior. "By exploring research in other fields, we can discover new ideas and methods to apply in our own work," Jafari shares. "My advice to students is to stay curious and keep exploring innovations beyond the classroom."