Presentation of classification model of occupational exposure to noise and heat based on multilayer perceptron neural network

Exposure to heat and noise can have negative impacts on cognitive and behavioral performance in various work and non-work settings. Therefore, this study was conducted with the aim of providing a model based on a multilayer perceptron neural network for occupational exposure classification of different levels of noise and heat.

This study involved the examination of 72 voluntary students, aged between 23 and 33 years. The input data provided to the neural network included cognitive, behavioral, physiological, and neurophysiological information. The output layer of the network was designed to classify exposure to various levels of noise and heat into three categories: lower, higher, and within permissible limits.

In this study, a two-layer neural network (15:10) was considered as the optimal model, with a chance of approximately 33 percent for correctly classifying the data. To evaluate this model, accuracy percentage, mean squared error (MSE), and sensitivity were calculated. The classification accuracy for different levels of noise and heat during the learning phase was 93.87 percent, and during the testing phase was 92.62 percent and the validity of 92.68 percent. Additionally, the mean squared error percentage was 0.53, and the sensitivity percentage was 90.42.

The present study demonstrated that the proposed model based on a multi-layer perceptron neural network has an acceptable accuracy and sensitivity for predicting different classes of noise and heat using psychophysiological and neurophysiological input data.