This review paper considers using machine learning (ML) methods in supply chain disruption prediction for the automotive sector. Due to the growing relevance of global events disruptive to the supply chain, timely and accurate disruption prediction has become pivotal in ensuring that production costs are minimal. Subsequently, we explore the different Machine Learning methodologies of supervised, unsupervised, and reinforcement learning and discuss their applicability for disruption prediction. Some major problems including data accessibility and quality, data privacy and model explainability are discussed in detail. As with other aspects of the implementation of data-driven methodologies, this review also provides recommendations concerning information storage and treatment as well as suggested algorithms for use, grounded in the proven examples of effective applications of ML. Finally, we determine further research prospects, stating the demand for real-time prediction, AI ethical concerns, and the connection of external databases for a higher level of model performance. Finally, this paper will educate the stakeholders on how they can use machine learning to manage different forms of supply chain disruption challenges.