This review paper examines the chronological progression of machine learning applications into structural health monitoring (SHM) for RCC and steel bridges over the past few decades. Key developments in sensing technologies, data processing methods, and machine learning algorithms are analyzed, highlighting the evolution from traditional inspection approaches to data driven predictive maintenance strategies. The review covers major milestones from early applications of artificial neural networks to recent implementations of deep learning and digital twin models. Challenges in data availability, model transferability, and long-term reliability are discussed, along with future research directions including physics guided machine learning and integration with emerging technologies. This comprehensive overview demonstrates how machine learning has significantly enhanced damage detection, condition assessment, and maintenance planning capabilities for bridge infrastructure.