Developing an Ensemble Predictive Safety Risk Assessment Model: Case of Malaysian Construction Projects.

Area of Science:

• Construction Management
• Occupational Health and Safety
• Machine Learning Applications

Background:

• Occupational Health and Safety (OHS) injuries pose significant challenges in developing countries' construction projects, often stemming from inadequate management, governmental oversight, and technical safety practices.
• Existing research on OHS in developing nations is limited, highlighting a need for advanced predictive assessment frameworks, particularly in rapidly growing sectors like Malaysia's construction industry.
• No prior studies have specifically addressed OHS risk assessment for construction workers in Malaysia, indicating a critical research gap.

Purpose of the Study:

• To develop and validate an Ensemble Predictive Safety Risk Assessment Model (EPSRAM) for assessing OHS risks faced by construction workers.
• To identify key safety risks and critical factors influencing OHS assessment in construction.
• To accurately predict the magnitude of safety risks and propose effective mitigation strategies.

Main Methods:

• Development of the Ensemble Predictive Safety Risk Assessment Model (EPSRAM) by integrating neural networks with fuzzy inference systems.
• Application and testing of the EPSRAM model on various construction projects within Malaysia.
• Utilizing machine learning techniques for predictive risk assessment and evaluation.

Main Results:

• The study successfully identified major potential safety risks prevalent in construction sites.
• Crucial factors affecting safety assessments for construction workers were determined.
• EPSRAM demonstrated accurate prediction of the magnitude of identified safety risks and applicable evaluation strategies.
• The model provides valuable insights for safety professionals and inspectors.

Conclusions:

• The developed EPSRAM serves as an effective tool for assessing OHS risks in construction, particularly in the Malaysian context.
• EPSRAM's predictive capabilities can significantly aid in improving the working environment and ensuring the well-being of construction workers.
• This research contributes a novel machine learning-based framework to the under-researched area of construction OHS in developing countries.