Data-driven machine learning methods, by establishing complex mapping relationships between material characteristic parameters and target properties, provide a novel paradigm for ductile iron research and development, overcoming the limitations of time and cost inherent in traditional research and development (R&D) approaches. Recent progress in the application of machine learning within the ductile iron R&D process is systematically reviewed. The fundamental implementation framework is elucidated, encompassing data collection, data preprocessing, model construction and training, and model evaluation. The applications of machine learning are summarized in various areas, including microstructure and defect control, prediction of mechanical properties, and prediction of service performance. Critical challenges demanding urgent solutions in machine learning-based ductile iron R&D and applications are discussed. Finally, research directions and future development trends for machine learning-driven ductile iron development are proposed.