Reducing serious injuries and fatalities in industrial construction-application of machine learning to analyze emotional intelligence and psychosocial factors

Human and organizational factors are recognized as central in incidents; however, there has been little interconnection between individual and organizational psychological variables, such as interpersonal skills. Therefore, this study aims to glean theoretical and empirical insights to reduce severe injuries and fatalities and enhance safety performance. This study classified incidents of an industrial construction organization in Canada based on two attributes of interpersonal skills–emotional intelligence (EI) and psychosocial (PS) factors. A qualitative analysis using NVivo software was employed to classify 1000 incidents from 2018 to 2020 into EI factors. Since PS factors were not observed in the dataset, the analysis was extended to identify PS factors using machine learning techniques as a quantitative approach to analyze 45,603 incidents from 2014 to 2023. The classification was performed using keyword analysis of the incident descriptions. Further, co-occurrence networks were used to investigate patterns and validate the study results. The findings indicate that lack of self-awareness (domain of EI) (56.8%) and improper communication (domain of PS factor) (32.4%) were the most influential causes of incidents substantiated by the co-occurrence networks results. The study's findings provide insights for decision-makers about the strategies needed to enhance safety performance in the industrial construction industry.