Artificial intelligence driven systems for enhancing worker safety monitoring in construction environments: a comprehensive review

Abstract

The construction industry remains one of the most hazardous sectors globally, largely due to inefficiencies, high operational costs, and delayed risk detection associated with traditional safety procedures. This comprehensive review synthesizes findings from 88 peer-reviewed studies published between 2017 and 2025 to evaluate the potential of Artificial Intelligence (AI)-driven systems in enhancing worker safety. The review addresses three primary objectives: (1) evaluating the effectiveness of AI technologies in improving safety monitoring, (2) examining regional variations in AI adoption and their influence on safety outcomes, and (3) identifying key challenges related to implementation, scalability, and ethical considerations. The analysis reveals that AI technologies significantly contribute to improved safety outcomes. For instance, drone-based inspections reduce assessment times from several days to approximately 1.5 hours with high accuracy; computer vision systems detect personal protective equipment (PPE) violations with over 91% precision; and predictive analytics forecast accident risks among migrant workers with 89% accuracy. Regional case studies highlight disparities, such as a 60% reduction in fall incidents in the United Arab Emirates through convolutional neural network (CNN) based surveillance, contrasted with barriers in Ghana, including high implementation costs and cultural resistance. Despite these advancements, challenges persist, including privacy concerns, technological limitations, and inadequate infrastructure in low-resource settings. The review underscores the need for explainable AI models, adaptable frameworks for diverse environments, and harmonized international regulations to ensure equitable and effective deployment of AI-driven safety systems.