Network-based safety risk analysis and interactive dashboard for root cause identification in construction accident management

Abstract

Construction projects are characterized by dynamic environments where numerous safety risks interact, leading to high rates of accidents and fatalities. Traditional safety analysis methods often overlook these complex relationships, hindering effective risk mitigation strategies. This study uses graph theory and network analysis to analyze the interconnectivity of safety factors in construction incidents. Using a dataset of injury and fatal accident cases, a network was constructed to represent safety factors and their co-occurrences. Centrality measures were applied to identify influential factors, while the Louvain algorithm facilitated community detection. The results identified PM₁₀ groups (air quality) and temporal factors (specific times of day) as key risks. Three major clusters of safety factors were also detected, representing environmental, incident-related, and demographic influences. An interactive dashboard was developed for scenario simulation, allowing construction professionals to visualize the effects of removing key factors from the network. These findings offer a practical framework for targeted safety interventions and real-time management of construction risks. The study concludes that integrating graph theory into construction safety analysis can provide a more comprehensive approach to accident prevention by focusing on interconnected risk factors rather than isolated incidents.