Analysis Of Correlation Between Window Duration For Kurtosis Computation And Accuracy Of Noise-induced Hearing Loss Prediction

Noise-induced hearing loss（NIHL）is one of the most common occupational related health problems worldwide,usually leading to long-term defects in language and cognitive development,understanding,behavior,and social adaptation.Studies have shown that kurtosis can effectively evaluate the biological effects of complex industrial noise and that it is a potential auxiliary index for hearing loss evaluation.However,the window duration has a large influence on the calculated kurtosis values:different kurtosis values are obtained when using different window durations,and a method for selecting a proper window duration for kurtosis computation has not been studied thoroughly.In this study,two algorithms,a statistical comparison（N=1,147）and a support vector machine NIHL prediction model（N=847）,were designed to investigate the correlation between window length and the accuracy of NIHL prediction by using a large human database collected from 22 industry in Zhejiang.In the statistical comparison,subjects were divided into high and low kurtosis groups based on kurtosis values computed over different window durations.The Mann-Whitney U test and Cohen’s d were used to compare the difference in HTLs between each two-group pair to find the optimal window duration to best distinguish these two groups.In the hearing loss prediction model,the kurtosis metric obtained from different window durations,together with age,exposure duration and L_Aeq,was used as a feature of the model for NIHL evaluation.The area under the curve was used to evaluate the performances of models to compare the influence of the window length on the evaluation of hearing loss.The results show that the kurtosis value based on most time windows has a significant correlation with noise-induced hearing loss,and the kurtosis value based on a relatively short time window（40-s to 2-min）is a better choice.Combining the two methods,60-s was an optimal window duration for kurtosis calculation that allows for both efficient computation and high accuracy for NIHL evaluation.The results presented in this study have significance for guiding future research on occupational noise-induced hearing loss evaluation using kurtosis.