A Comprehensive Review of Machine Learning and Explainable AI Approaches for Road Accident Severity Prediction

Road traffic accidents continue to be one of the major causes of injuries and fatalities across the globe, leading to severe social and economic consequences. In recent years, advancements in artificial intelligence have encouraged the use of machine learning (ML) and deep learning techniques to analyze accident data and predict accident severity. A variety of predictive approaches, including Random Forest, Neural Networks, Support Vector Machines, and deep learning architectures, have been explored to improve accident severity classification. Despite these developments, many existing studies face challenges such as limited dataset sizes, poor model interpretability, class imbalance problems, and lack of real-time implementation. Moreover, the transparency of predictive models has become an essential requirement for their practical use in transportation safety systems. This study provides a comprehensive review of fifteen recent research works focusing on accident severity prediction, hotspot identification, driver behavior analysis, and explainable artificial intelligence (XAI). The research compares different methodologies, datasets, and performance outcomes in order to highlight current research trends and identify gaps in the literature. Based on this analysis, a conceptual framework integrating machine learning, deep learning, and explainable AI techniques is proposed to enhance prediction accuracy and model interpretability. The proposed approach aims to support intelligent transportation systems and assist decision-makers in improving road safety strategies.