LAUSR

Abstract Construction safety prediction is an emerging field where various forms of information and analytical techniques are used to predict the likelihood or severity of a future injury. A review of this literature reveals that even though the approaches are used for the same goal of predicting future safety outcomes, they are modeled independently and exclusively from one another. To organize thinking in safety prediction, the literature is organized into four operationally-defined predictive families: (1) safety risk assessment, which considers the characteristics and dangers of the work; (2) precursor analysis, which considers the conditions of the workers; (3) leading indicators, which consider the quantity of safety management activities; and (4) safety climate assessments, which considers worker perceptions of safety. Additionally, a unified model is proposed where the four families are considered together and opportunities for synergy and cross-validation are exploited. Researchers may benefit from this model as they create points of departure, propose and test novel approaches, and attempt to contextualize their findings within the existing body of literature. Furthermore, practitioners may use the model to make more accurate and robust safety predictions that account for the interconnectedness of the work attributes, human resources, and management strategies that affect safety.