Development of noise induced hearing loss prediction model using artificial neural network / Siti Fairus Mohd Zain

Noise Induced Hearing Loss (NIHL) was the highest reported cases of occupational disease in 2016. Despite the high incidence reported, studies in the method of predictive modelling causes were limited. Hence, this research proposed the development of Artificial Neural Network (ANN) as a tool to identify and predict risk factors contributed to NIHL. ANN was chosen in this study since it was proven to predict few diseases including coronary heart disease, diabetes, liver cancer and otitis media disease. There are a lot of prediction techniques available in computational models, but this project explored on the Feed Forward Backpropagation Networks as it has been used in predicting complex diseases. This model using a design approach of 24 inputs and 5 binary output layers. The 24 input layers encompassed 12 risk factors and 12 audiogram variables. It also embedded with 10 hidden layers in the prediction models using Levenberg-Marquardt algorithm as a transfer function from input vectors to the five binary outputs. The binary output vectors referred are according to the World Health Organization (WHO) standard, which are classified as either normal, mild, moderate, severe, and profound. The study was focus on examining 355 secondary data extracted from NIHL confirmed cases provided by the Department of Occupational Safety and Health (DOSH), Selangor State.