This systematic review, using the PRISMA framework, provides a state-of-the-art synthesis of Artificial Intelligence (AI) in bridge Structural Health Monitoring (SHM), highlighting its transformative potential in redefining how structural integrity is assessed. From 99 selected studies, a clear paradigm shift has emerged toward deep learning architectures as the dominant analytical paradigm, which is increasingly employed for critical SHM tasks such as automated damage detection, intelligent data management, predictive maintenance modeling, and structural system identification. AI-driven approaches consistently demonstrate robust solutions to the longstanding drawbacks of traditional SHM methods, manual inspections that are subjective, labor-intensive, and financially demanding. However, despite these promising developments, persistent challenges continue to hinder the field-scale deployment of AI-enabled SHM systems, particularly data scarcity, severe class imbalance, and the disruptive influence of Environmental and Operational Variability (EOV) on monitoring accuracy. Looking ahead, the literature advocates integrating multi-modal sensing data and advancing explainable, physics-informed AI models. Combining these with digital twins can yield the next generation of robust, field-ready SHM systems for bridges.
Ataei et al. (Mon,) studied this question.