ANALYSIS OF VEHICLE PEDESTRIAN CRASH SEVERITY USING ADVANCED MACHINE LEARNING TECHNIQUES

Aug 3, 2026

Vehicle-pedestrian collisions remain a major road-safety challenge, with pedestrians representing a large share of fatalities in Hong Kong across multiple years. Conventional crash-severity studies often overlook within-crash correlations between pedestrian-vehicle components, which can bias factor-effect estimates. To address these limitations, the study applies a Bayesian neural network (BNN) and compares it with K-Nearest Neighbors, a standard artificial neural network, and random forest models. Results show BNN yields superior performance and that accident timing, circumstances, and meteorological features meaningfully improve predictions to support fatality reduction.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages27

Reading Time1 h 8 min

AccessFree to download

Introduction

Background on pedestrian vulnerability and walkability

Limitations of traditional crash-severity modeling

Methodology

Bayesian neural network approach

Comparative machine learning models

Results and Discussion

Model performance and predictive accuracy

Importance of temporal, situational, and meteorological factors

Conclusion and Implications

Framework for reducing pedestrian fatalities

ANALYSIS OF VEHICLE PEDESTRIAN CRASH SEVERITY USING ADVANCED MACHINE LEARNING TECHNIQUES Siyab UL ARIFEEN1, Mujahid ALI2, Elżbieta MACIOSZEK3 1 Department of Civil Engineering, COMSATS University Islamabad, Abbottabad, Pakistan 2, 3 Department of Transport Systems, Traffic Engineering and Logistics, Faculty of Transport and Aviation Engineering, Silesian University of Technology, Katowice Abstract: In 2015, over 17% of pedestrians were killed during vehicle crashes in Hong Kong while it raised to 18% from 2017 to 2019 and expected to be 25% in the upcoming decade. In Hong Kong, buses and the metro are used for 89% of trips, and walking has traditionally been the primary way to use public transportation. This susceptibility of pedestrians to road crashes conflicts with sustainable transportation objectives. Most studies on crash severity ignored the severity correlations between pedestrian-vehicle units engaged in the same impacts. The estimates of the factor effects will be skewed in models that do not consider these within-crash correlations. Pedestrians made up 17% of the 20,381 traffic fatalities in which 66% of the fatalities on the highways were pedestrians. The motivation of this study is to examine the elements that pedestrian injuries on highways and build on safety for these endangered users. A traditional statistical model's ability to handle misfits, missing or noisy data, and strict presumptions has been questioned. The reasons for pedestrian injuries are typically explained using these models. To overcome these constraints, this study used a sophisticated machine learning technique called a Bayesian neural network (BNN), which combines the benefits of neural networks and Bayesian theory. The best construction model out of several constructed models was finally selected. It was discovered that the BNN model outperformed other machine learning techniques like K-Nearest Neighbors, a conventional neural network (NN), and a random forest (RF) model in terms of performance and predictions. The study also discovered that the time and circumstances of the accident and meteorological features were critical and significantly enhanced model performance when incorporated as input. To minimize the number of pedestrian fatalities due to traffic accidents, this research anticipates employing machine learning (ML) techniques. Besides, this study sets the framework for applying machine learning techniques to reduce the number of pedestrian fatalities brought on by auto accidents. Keywords: Machine learning, ANN, BNN, Vehicle-pedestrian crash To cite this article: Ul Arifeen, S., Ali, M., Macioszek, E., (2023). Analysis of vehicle pedestrian crash severity using advanced machine learning techniques. Archives of Transport, 68(4), 91-116. DOI: [https://doi.org/10.61089/aot2023.ttb8p367](https://doi.org/10.61089/aot2023.ttb8p367) Contact: 1) [siyabularifeen@cuiatd.edu.pk](siyabularifeen@cuiatd.edu.pk) [[https://orcid.org/0009-0001-3785-9404](https://orcid.org/0009-0001-3785-9404)] 2) [mali@polsl.pl](mali@polsl.pl) [[https://orcid.org/0000-0003-4376-0459](https://orcid.org/0000-0003-4376-0459)] – corresponding author; 3) [elzbieta.macioszek@polsl.pl](elzbieta.macioszek@polsl.pl) [[https://orcid.org/0000-0002-1345-0022](https://orcid.org/0000-0002-1345-0022)] 1. Introduction Walking is a vital part of daily transportation since it is a physically and environmentally healthy mode of travel, especially for first and last-mile movement. However, due to their lack of protection, pedestrians or foot travelers are the most venerable road users who are more likely to sustain severe injuries or die in car accidents (Behnood and Mannering, 2016, Pucher and Dijkstra, 2003, Liu et al., 2019, Hafeez et al., 2023) . Significant socioeconomic repercussions result from pedestrian mortality and injuries in traffic accidents. This is especially important when taking into account the ongoing initiatives taken by developed econom

FAQs

Why do researchers focus on pedestrian crash severity in vehicle-pedestrian collisions?

Pedestrians are highly vulnerable road users due to limited protection, and pedestrian mortality and injuries create significant social and safety consequences. The study targets severity to support more effective prevention.

What modeling limitation is addressed in this study?

Many prior studies ignore correlations between severity components within the same crash involving pedestrian-vehicle units, which can skew factor-effect estimates in models that assume strict conditions and handle missing or noisy data poorly.

Which machine learning model performed best and what key inputs improved it?

The Bayesian neural network (BNN) outperformed K-Nearest Neighbors, a conventional neural network, and a random forest in performance and predictions. Accident timing and circumstances, together with meteorological features, significantly enhanced model performance when used as inputs.

Aug 3, 2026

Upload by Olivia Brown