Improving Safety through Leveraging Machine Learning and Safety-Related Data in the Construction Industry

Abstract This study presents a conceptual framework for integrating safety-related data with machine learning to improve its understanding of safety performance and construction safety management. Machine Learning techniques could discover latent hazards and risks by utilizing project-specific safety-related data and improve safety and decision-making processes. Findings suggest that machine learning can significantly improve safety performance by proactively identifying risks and measures from safety-related data rather than relying solely on historical safety outcomes and data. This could also provide a better understanding of the forthcoming construction projects’ complex challenges and the impact of increasingly technical and organizational complexities on safety. However, challenges such as data compatibility, lack of standardization, misinformation risks, and ethical concerns (transparency, privacy, and fairness) necessitate a cautious approach to the use of machine learning. This proactive approach could lead to safer construction environments and continuous improvements in safety management. Future work will refine data collection and develop predictive models, with the current research in the ‘DiSCo’ project aiming for sustainable safety improvements in the construction industry.