Modeling and Weighting of Factors Affecting Sustained Attention and Working Memory of Workers Exposed to Different Sound Pressure Levels using Deep Learning and Random Forest Algorithms: A Case Study of a Steel Industry

Abstract

Introduction: Cognitive functions play a vital role in how tasks are performed; for this, temporary cognitive and mental dysfunctions could lead to grave consequences, especially when an accurate and prompt response is required. Attention and reaction time to noise are among the most effective exogenous factors on the brain processing mechanism. This study aimed to measure the sustained attention of workers in the steel industry exposed to different sound pressure levels. Material and Methods: The study was conducted in 4 general stages, including 1- Selecting predictive orientation variables (age, work history, different sound pressure levels); 2- Conducting the Cognitive Performance Test (CPT); 3 Conducting N-BACK Cognitive Performance Test and 4- Modeling cognitive performance changes using model precision methods. Results: Continuous Performance Test (CPT) results indicated that all three groups’ omission error, commission error, and response time were affected by shift time. All three components increased significantly as the shift ended, decreasing individuals’ cognitive function. Also, the higher noise impact in modeling CPT and N-Back tests indicated reduced workers’ concentration. Conclusion: These study findings suggested that greater noise weight obtained in test modeling in three-time intervals, i.e., in the beginning, middle, and end of the shift, affected the continuous performance components of the CPT and working memory performance of the N-back test, including workers’ response time and reaction time, with workers’ rate of error increasing and their focus decreasing during the shift.

Keywords

• modeling
• sustained attention
• deep learning
• random forest
• noise