Emotional responses play a crucial role in decision-making processes and human interactions. This study used the FaceReader technology to investigate the correlation between emotional expressions (fear, anger, happiness, sadness, disgust, surprise and arousal) and hazard recognition performance. The facial expressions of the participants during the hazard identification test were recorded and then analyzed using machine learning methodologies. The results highlighted the significant relationship between emotional response and hazard identification performance, with heightened anger and happiness indicating higher hazard identification scores, while increased arousal and fear was associated with decreased performance. These findings demonstrate the relationship between emotional states and hazard identification performance. This study highlights the potential of integrating FaceReader technology with machine learning tools to predict the safety behavior of employees in the workplace and prevent hazards before they occur and offers a data-driven approach using objective measures in improving workplace safety.