Authors: Edoardo Maria Polo12, Marta Lenatti2, Marco Zanet2, Riccardo Barbieri2, Alessia Paglialonga3
1DIAG, Sapienza University of Rome, Italy
2Politecnico di Milano, Dipartimento di Elettronica, Informazione e Bioingegneria (DEIB), Milan, Italy
3National Research Council of Italy (CNR), Institute of Electronics, Computer and Telecommunication Engineering (IEIIT), Milan, Italy
Background: Individuals with hearing loss usually seek help very late, when major effects have already occurred. In this context, speech-in-noise screening tests can give individuals with hearing problems an initial chance to assess their hearing quickly, both in situ and in remote settings (e.g. at home). The aim of this study was to evaluate the ability of a recently developed automated, language-independent speech-in-noise screening test to predict hearing loss in adults. In particular, we investigated the relationship between the Speech Reception Threshold (SRT) measured by the newly developed test and the Pure Threshold Average (PTA), defined as the average of pure-tone hearing thresholds measured at 0.5, 1, 2, and 4 kHz.
Methods: Two groups of subjects were tested: 36 normal hearing young adults (age: 20 to 26 years, mean age: 23.8 years; Nsubjects = 36; Nears = 36) and 114 unscreened adults (age: 12 to 89 years, mean age: 60.3 years; Nsubjects = 114; Nears = 122). Receiver operating characteristic (ROC) curves were computed to characterize the speech-in-noise test against the WHO criteria for “slight/mild” (criterion (1): PTA > 25 dB HL) and for “moderate” hearing loss (criterion (2): PTA > 40 dB HL). Generalized linear models (GLM) were computed to assess the SRT and age as predictors of the grade of hearing loss, as defined by the two WHO criteria.
Results: The ROC curves for the two WHO criteria show a satisfactory performance of the proposed test in predicting PTA outcomes. Indeed, the area under the curves are equal to 0.83 (SRTcut-off = -8.87 dB SNR, Sensitivity = 74%, Specificity = 84%, Accuracy = 81%) for criterion (1) and to 0.89 for criterion (2) (SRTcut-off = – 6 dB SNR, Sensitivity = 89%, Specificity = 81%, Accuracy = 82%).
For each of the two WHO criteria, the SRT in a single-predictor GLM shows a significant correlation (p << 0.01 for both criteria). Regarding the GLM with SRT and age, both predictors are significantly correlated with PTA outcomes for criterion (1) (SRT: p = 0.006, Age: p << 0.01) whereas only SRT is a significant predictor of PTA outcomes for criterion (2) (SRT: p << 0.01, Age: p = 0.09).
Conclusion: The proposed language-independent speech-in-noise test showed a good performance in terms of predicting PTA outcomes, particularly for the identification of moderate hearing loss. Further research is needed to fully validate the test in a larger population and to investigate the relationship between SRT and PTA and possible additional factors that may predict PTA outcomes.