Unveiling Construction Accident Causation: A Scientometric Analysis and Qualitative Review of Research Trends

Haoyu Zang^*Ming LiZhiyao JinJingfei Huang

• School of Management Engineering, Qingdao University of Technology, Qingdao, China

The construction industry, a cornerstone of global economic growth, faces frequent safety accidents due to its complex environments and multi-party collaboration, impeding sustainable development. These incidents arise from interlinked causal factors, including human error, management shortcomings, technical failures, and environmental conditions. This study systematically reviews construction accident causation research by integrating scientometric analysis and qualitative methods, using VOSviewer to analyze literature from Scopus and Web of Science databases, with 110 peer-reviewed articles selected through a validated Boolean search strategy. VOSviewer was used for bibliometric visualization to map research trends, co-authorship networks, and keyword cooccurrences. In addition, a qualitative synthesis was conducted to review common data sources and examine key issues, including risk factor identification, accident type classification, causality analysis, and the optimization of research strategies. The study aims to systematically review the current state of construction accident causation research, highlighting key trends in data-driven and AI-based safety interventions. Findings reveal a shift toward data-driven, intelligent approaches, with artificial intelligence techniques-such as large models (capable of understanding complex patterns from massive datasets), graph neural networks (suitable for modeling relationships between contributing factors), and natural language processing (for extracting insights from textual accident reports)-enhancing accident prevention and risk prediction. Challenges persist, however, in data quality, causal exploration depth, and interdisciplinary integration. These findings underscore the need for further advancements in data accuracy and model scalability, which could inform more effective safety management practices and policy frameworks. Key contributions include filling the bibliometric gap in this field, offering a novel framework combining quantitative and qualitative insights, and highlighting advanced technology applications, thus providing theoretical and practical guidance for future safety management. Future research is recommended to leverage AI, foster interdisciplinary collaboration, and develop precise prevention systems to address these gaps.