Research On Complex Noise Induced Hearing Loss Prediction Model Based On Machine Learning

Hearing loss is a major public health problem facing the world.The main cause of hearing loss is noise exposure.Therefore,fully exploiting the potential value contained in the noise exposure big data to accurately estimate the noise-induced hearing loss plays an important role in saving medical costs and enhancing hearing protection.The current standard for assessing noise-induced hearing loss is the International Noise Exposure Standard(ISO-1999).However,the establishment of this standard is based on the steady-state noise data of the 1950s and 1960s,so it is not sensitive to the type of noise exposure,and often underestimates the hearing loss caused by complex noise.In addition,the inability to utilize other effective characteristic parameters other than energy is another important reason for limiting the standard's accurate prediction of biological effects caused by complex noise.Data-driven machine learning methods can make full use of the effective information contained in the data,so it is of great significant to establish a more accurate prediction model of hearing loss based on noise exposure big data and machine learning algorithms to assess the impact of noise exposure on hearing impairment.This paper studies the new methods of hearing loss prediction under complex noise exposure from the perspective of machine learning,and improves the performance of hearing impairment prediction.The model has a reasonable interpretation of audiology and is of great significance for hearing loss prevention and early intervention.The main innovations are summarized as follows:Aiming at the prediction of hearing loss caused by complex noise,this paper proposes machine learning methods based on SVM,MLP,random forest and AdaBoost,and effectively utilizes the potential information in complex noise data to establish a comprehensive prediction model of hearing loss and damage,and the effects of various machine learning algorithms are compared systematically.A fusion method of complex noise characteristics and individual features is proposed to solve the insensitivity of traditional methods to noise exposure types,a hearing impairment prediction model is established,and the prediction performance is significantly better than the existing industry international standardsA feature discovery method based on convolutional neural network for complex noise time series model is proposed.The time patterns existing in the noise time series data processed into two-dimensional matrix are captured,and a more accurate prediction model of hearing damage caused by complex noise is established.