Conceptual Model for Predicting Worker Safety Behaviour in Construction Projects: A Neural Network Approach
Currently, fatality statistics as indicated by Safe Work Australia report that 35 construction workers are seriously injured each day on construction sites. Work-related injuries, illnesses and deaths impose costs on employers, workers and the community. To the employer, these costs constitute overtime premium, employer access payments, sick leave, staff turnover costs, threshold medical payments, real legal costs incurred plus fines and penalties, and/or employer investigation costs. Reducing the number, rates, and costs of accidents, incidents, and injuries is an ongoing challenge for researchers, policymakers, and practitioners. It is therefore essential for managers to be aware of the likely safety performance on a project and to take any necessary remedial measures to reduce the likelihood of accidents.
Currently, the use of data mining techniques such as artificial neural networks in other industries such as energy and electronics have led to the discovery of knowledge from data for effective decision making. Results from these techniques have been known to be promising. Policy makers, construction organisations, and consultancies would be interested to employ the conceptual model as the basis to effectively solve fuzzy and complex problems.
Key ‘take aways’ or ‘learning outcomes:
- The development of artificial neural network (ANN) problems provides the opportunity and possibility to establish an unsafe behaviour monitoring system.
- Despite the importance of safety climate in the promotion of construction safety, previous studies have not been focused on how to apply safety climate to accident prediction. As a leading indicator or precursor, safety climate has the great potential to timely predict accidents on construction sites. This study illustrates how to use safety climate for prediction.
- Understanding the future trend of safety performance could provide insight for developing appropriate safety policies and strategies to mitigate the likelihood of accidents.
- An effective and supportive environment may be developed when management can foresee the likely adverse state of safety on construction projects.
- The model could serve as the theoretical basis for construction organisations to monitor, evaluate, predict, and encourage compliance with current processes and design better interventions.
- Due to practical nature applying ANN, researchers and practitioners can work cooperatively and explore the effective path of machine learning transition from construction safety research into construction safety practice.