Supply chain disruptions pose significant challenges to global economic stability, necessitating advanced predictive tools for effective risk management. As Machine Learning (ML) offers promising solutions for enhancing resiliency, this study investigates its applications in supply chain management. Utilizing a systematic literature review, we examined recent research to identify effective ML models and techniques, focusing on both supervised and unsupervised learning. Our analysis covered various industries to understand the adaptability and effectiveness of these models in mitigating supply chain risks. The results highlight the growing implementation of ML in anticipating disruptions, with supervised learning demonstrating superior predictive precision under specific conditions. At the same time, unsupervised approaches offer valuable insights in data-scarce scenarios. Context-specific data surfaced as crucial in model accuracy, underscoring the need for tailored approaches. This study concludes that integrating ML with current supply chain systems can significantly enhance operational resilience, advocating for continued exploration of novel data sources and interdisciplinary collaborative efforts.