Related Experiment Video
Updated: Jul 11, 2025
08:05
Design and Analysis for Fall Detection System Simplification
Published on: April 6, 2020
10.7K
Advances, challenges, and future research needs in machine learning-based crash prediction models: A systematic
Yasir Ali1, Fizza Hussain2, Md Mazharul Haque2
1School of Architecture, Building, and Civil Engineering, Loughborough University, Leicestershire LE11 3TU, United Kingdom.
Accident; Analysis and Prevention
|November 17, 2023
Summary
This review summarizes machine learning models for predicting road crashes and injury severity. It identifies gaps and future research needs for improving road safety through advanced crash prediction.
Area of Science:
- Road safety
- Traffic engineering
- Data science
Background:
- Accurate crash modeling is crucial for effective road safety management and developing countermeasures.
- Machine learning (ML) has been applied to crash prediction for over 20 years, with big data offering new opportunities.
- Understanding state-of-the-art ML crash prediction models is essential for future advancements.
Approach:
- This paper systematically reviews ML studies on crash modeling, categorizing them by application: crash occurrence, crash frequency, and injury severity prediction.
- The review analyzes model intricacies affecting performance and identifies specific research gaps and needs.
- Methodological advancements and the use of big data in ML crash prediction are discussed.
Key Points:
- Current crash occurrence models struggle with selecting appropriate non-crash events.
- Crash frequency models lack the ability for future forecasting.
- Inconsistent injury severity classifications pose a challenge in current models.
Conclusions:
- This review highlights critical research needs in ML model development, evaluation, and application for road safety.
- Future research should focus on addressing identified gaps to develop more robust and accurate crash prediction models.
- Leveraging big data and advancing ML methodologies can significantly improve the understanding and prediction of road crashes and their determinants.
