The development of models for the prediction of noise-induced hearing loss.
Qiu-W; Ye-J; Hamernik-RP
J Acoust Soc Am 2006 Nov; 120(5)(Part 2):3128
Three statistical learning models were developed to predict noiseinduced hearing loss (NIHL) from an archive of animal noise exposure data, which contains 936 chinchillas exposed to various noise environments. The following models were constructed: (i) A support vector machine model with a nonlinear radial basis function kernel. (ii) A multilayer perceptron network model and (iii) a radial basis function network model. In addition to frequency-specific energy metrics, noise exposure parameters and biological metrics such as kurtosis, noise type, and pre/ postexposure hearing thresholds were used as inputs to the model. There were several indices of auditory trauma at specific audiometric test frequencies that were to be predicted by the models: e.g., noise-induced permanent threshold shift, percent outer hair cell loss, and percent inner hair cell loss. The average prediction accuracy for the three models was better than 80%. These results demonstrate the feasibility of developing such models for the prediction of NIHL in humans.
Noise-exposure; Noise-induced-hearing-loss; Hearing-loss; Hearing-impairment; Hearing-disorders; Models; Animal-studies; Laboratory-animals
Wei Qiu, PhD, Auditory Research Laboratory, State University of New York at Plattsburgh, 101 Broad Street, Plattsburgh, New York, 12901
Disease and Injury: Hearing Loss
Journal of the Acoustical Society of America