AI-ENABLED FIRE DETECTION AND PREDICTIVE SAFETY MONITORING FOR SMART BUILDINGS IN SAUDI ARABIA UNDER VISION 2030: A STATE-OF-THE-ART REVIEW AND FRAMEWORK

Fire safety in smart buildings is moving from isolated threshold alarms towards data-rich systems that can recognise weak precursors, estimate fire development and support coordinated response. This state-of-the-art review synthesises recent peer-reviewed evidence on artificial intelligence,computer vision, multimodal sensing, building information modelling, digital twins, Internet of Things architectures and predictive fault monitoring, with specific interpretation for Saudi smart buildings. The evidence indicates that deep-learning models can improve visual smoke and flame recognition, while multimodal sensing and data-driven fire-state estimation can add information on location, intensity and likely evolution. Their safety value, however, depends on resilient sensing,edge processing, verified building context, calibrated uncertainty, cybersecurity and human oversight. The paper proposes a layered architecture in which AI augments rather than replaces approved deterministic fire-alarm functions, and it embeds a lifecycle assurance loop for commissioning, monitoring and revalidation.Saudi deployment must also align with the applicable Saudi Building Code, Civil Defense approval and witnessing processes, relevant HCIS requirements for in-scope assets, and recognised fire-alarm documentation and inspection practices such as NFPA 72.