Authors: Ramón Hernández-Villoria¹
¹Centro Clinico de Audición y Lenguaje Cealca
Background: The calculations used to obtain a person’s hearing level through behavioral tonal audiometry are based on a simple arithmetic average of the threshold values at four frequencies (PTA-4). While this method served to resolve compensation claims for war veterans eighty years ago, even its creators were aware of its limitations. Today, these averages, with 25 decibels considered the upper limit of normal hearing, are still used in most parts of the world.
Methods: The transitions from linear arithmetic to multi-parametric algorithms were implemented in the R environment. We integrate weighted frequency variables, adjusting for age-correction and binaural summation effects. The model incorporates nonlinear regression and psychoacoustic weighting functions that account for spectral smearing and speech-in-noise interference. This R-based framework allows for the processing of raw threshold data into high-resolution functional profiles, moving beyond the static 25 dB cutoff.
Results: The implementation yields specific diagnostic ratios and spectral slope coefficients that demonstrate a higher correlation with speech perception deficits than the traditional PTA-4. Our results indicate that these parameters can identify subclinical hearing loss patterns and provide probabilistic diagnostic mappings. Specifically, the R-systematized ratios facilitate the differentiation between sensory and neural involvement through automated pattern recognition, offering a sensitivity increase in predicting functional disability compared to legacy averaging methods.
Conclusions: The integration of these advanced computational models marks a definitive departure from the “pencil-and-paper” paradigm. By systematizing in R, we provide a scalable, open-access framework. This approach not only empowers manufacturers to develop the next generation of “smart” audiometers but also democratizes precision diagnostics for clinicians worldwide.
