A Systematic Review on Road Accident Prediction: A Special Attention Toward Machine Learning and Deep Learning Approaches

Traffic crashes are a significant factor in causing fatalities and injuries. International reports have suggested that there are significant social and economic costs that follow an accident. In the current era, researchers have explored the usage of ML and DL methods to predict road accidents and improve road safety. This systematic literature review analyzes 78 papers on the topic of using machine learning and deep learning to forecast road accidents and improve road safety. The review focuses on accident prediction, accident severity prediction, ML algorithms, IoT integration and DL models. The study highlights the various approaches used by researchers to develop predictive models for road accidents and discusses the advantages and limitations of different ML and DL methods. The findings indicate that these methods have the potential to accurately predict road accidents, providing insights into the contributing factors and allowing for targeted interventions. The paper concludes by outlining future research directions, including the need for more comprehensive datasets, novel DL techniques and integration with traffic management systems. This review produces a comprehensive recap of the research progress to date on this important topic and highlights potential avenues for further exploration.