Sentence Recognition Prediction for Hearing-impaired Listeners in Stationary and Fluctuation Noise With FADE: Empowering the Attenuation and Distortion Concept by Plomp With a Quantitative Processing Model.

The study tested a computational model, not a treatment or clinical decision, so it cannot show that using FADE improves patient care. The best prediction used information from a separate noise measurement, which the authors say makes the comparison with SII unfair. The model used only basic hearing impairment assumptions: audiogram thresholding and one level uncertainty parameter; it ignored outer hair cell compression, temporal fine structure, brainstem effects, and binaural interaction. Listeners varied widely in age from 23 to 82 years, and the model did not include age or cognitive factors, which may affect speech recognition in noise. Even after individual correction, a large remaining variance across subjects was reported, so one parameter cannot capture all effects of ageing and hearing impairment. A full set of audiogram and level uncertainty combinations was not simulated because of computational complexity.

Do not treat the FADE model as a clinical test for hearing aid fitting or communication outcomes. The paper validates a computational prediction against existing German Matrix data, and the best accuracy used extra data from a second noise condition that the standard SII did not have.