Harnessing Data-mining Algorithms to Model and Evaluate Factors Influencing Distortion Product Otoacoustic Emission Variations in a Mining Industry

Zare, Sajad; Esmaeili, Reza; Nakhaei Pour, Mojtaba

doi:10.4103/ijehe.ijehe_20_24

Harnessing Data-mining Algorithms to Model and Evaluate Factors Influencing Distortion Product Otoacoustic Emission Variations in a Mining Industry

Document Type : Original Article

Authors

Sajad Zare ¹

Reza Esmaeili ²

Mojtaba Nakhaei pour ²

¹ Department of Occupational Health Engineering and Safety at Work, Faculty of Public Health, Kerman University of Medical Sciences, Kerman, Iran

² Department of Occupational Health and Safety Engineering, Student Research Committee, School of Health, Isfahan University of Medical Sciences, Isfahan, Iran

10.4103/ijehe.ijehe_20_24

Abstract

Aim:
The aim of this study was to gauge the condition of otoacoustic emissions (OAEs) in workers, followed by modeling and estimating the weight of factors affecting changes in their emissions.
Methods:
The present study comprises two main phases. In the first phase, the OAEs were assessed using the distortion product OAEs (DPOAEs) test. Furthermore, the occupational factors influencing fluctuations in OAEs, including sound exposure, frequency, age, work experience, and exposure time, were measured. In the second phase, the weight of the factors affecting OAEs was investigated using deep learning (DL) and support vector machine (SVM) algorithms.
Results:
The results of both algorithms showed that sound exposure had the greatest effect (weighting between 36% and 45%) on the changes in OAEs. Frequency, with a weight ranging from 19% to 25%, was recognized as the second factor impacting the changes in DPOAEs. Conversely, age had the slightest effect on OAEs (weighing between 6% and 11%). The results also showed that the DL algorithm had higher accuracy compared to the SVM algorithm.
Conclusions:
As a result of determining the weight of factors causing variations in OAEs, the allocation of resources for control measures and effective reduction will be accomplished more efficiently and accurately.

Keywords

Modeling

otoacoustic emissions

signal-to-noise ratio