Predicting Injury Severity in Traffic Accidents using Statistical and Machine Learning Models

DOI

10.2991/978-94-6239-693-7_89How to use a DOI?

Keywords

Traffic Crash Severity Prediction; Ensemble Machine Learning; Stacking Classifier; SHAP; Feature Importance; SMOTE

Abstract

Road traffic accidents cause hundreds of thousands of injuries, deaths, and economic costs annually, and hence the prediction of severity is significant for effective rescue resource allocation and policy intervention [1]. Machine learning (ML) has been shown to be a valuable tool in achieving the patterns in massive databases of road traffic accidents that are not visible to human observation. ML algorithms process different types of information such as weather, road conditions, traffic, time, and geographical position, which are all factors in the prediction of accident severity [7]. Ensemble learning algorithms like Random Forest, XGBoost, and LightGB are known to be particularly effective in leveraging the strengths of multiple decision trees to uncover complex relationships between variables [6]. The authors enhance these algorithms using stacking and automated parameter tuning. In addition to being accurate, it is essential for an algorithm to be interpretable. This is achieved by algorithms like SHAP (SHapley Additive exPlanations), which can reveal the variables that contribute most to the predictions [7]. Another issue is the problem of class imbalance because the more severe accidents are not as common, and this issue is handled by methods such as SMOTE (Synthetic Minority Oversampling Technique), which generates new samples of the minority but crucial class [11]. With the combination of ensemble learning, interpretability, and balancing, reliable and interpretable models for traffic crash severity prediction can be built, which can result in safer roads with reduced traffic injuries [12].

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Dev Pratap Singh
AU  - Anuj Sharma
AU  - Bharat Bharat
AU  - Aditya Singh
AU  - Nidhi Singh
PY  - 2026
DA  - 2026/06/16
TI  - Predicting Injury Severity in Traffic Accidents using Statistical and Machine Learning Models
BT  - Proceedings of the International Conference on Intelligent Systems for a Sustainable Future (ISSF 2026)
PB  - Atlantis Press
SP  - 923
EP  - 934
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6239-693-7_89
DO  - 10.2991/978-94-6239-693-7_89
ID  - Singh2026
ER  -