A Novel Method for Classifying Hearing Impairment in Epidemiological Studies of Aging: The Wisconsin Age-Related Hearing Impairment Classification Scale

Audiograms and Prevalence of Hearing Loss in U.S. Children and Adolescents 6-19 Years of Age

PURPOSE

The aim of this study was to determine median audiograms and prevalence of hearing loss for U.S. 6- to 19-year-olds.

METHOD

U.S. national audiometric data for 2,709 six- to 19-year-olds from the National Health and Nutrition Examination Survey 2017-2020 were analyzed. Analyses were primarily descriptive and relied on distribution-free medians, cumulative distribution functions (CDFs), and population-weighted prevalence estimates. Differences between CDFs were examined using Cohen's h effect size. Results are presented for the age range of 6-19 years. Main effects of sex, age, and race/ethnicity were examined.

RESULTS

Differences in the audiograms between the sexes and among the various race/ethnicity groups were minor, but age group differences were notable. Pure-tone thresholds at 500 and 8000 Hz for the 6- to 9-year age group were 3-5 dB higher (worse) than those of the other age groups. The overall prevalence of hearing loss among 6- to 19-year-olds, defined as either of the pure-tone averages (PTAs) for 500, 1000, and 2000 Hz (PTA512) or 3000, 4000, 6000, and 8000 Hz (PTA3468k) exceeding 15 dB HL in either ear, was 11.24% (95% confidence interval [10.00%, 12.48%]). Binary logistic regression analyses found the odds of hearing loss to be significantly higher for individuals from families with a low income as well as for those with abnormal otoscopy or abnormal tympanometry.

CONCLUSIONS

For children and adolescents in the United States, hearing was unaffected by differences in sex and race/ethnicity, with only age impacting pure-tone thresholds and only at the lowest (500 Hz) and highest (8000 Hz) frequencies examined. The overall prevalence of hearing loss was about 11%.

Easy as 1-2-3: Development and Evaluation of a Simple yet Valid Audiogram-Classification System

Almost since the inception of the modern-day electroacoustic audiometer a century ago the results of pure-tone audiometry have been characterized by an audiogram. For almost as many years, clinicians and researchers have sought ways to distill the volume and complexity of information on the audiogram. Commonly used approaches have made use of pure-tone averages (PTAs) for various frequency ranges with the PTA for 500, 1000, 2000 and 4000 Hz (PTA4) being the most widely used for the categorization of hearing loss severity. Here, a three-digit triad is proposed as a single-number summary of not only the severity, but also the configuration and bilateral symmetry of the hearing loss. Each digit in the triad ranges from 0 to 9, increasing as the level of the pure-tone hearing threshold level (HTL) increases from a range of optimal hearing (< 10 dB Hearing Level; HL) to complete hearing loss (≥ 90 dB HL). Each digit also represents a different frequency region of the audiogram proceeding from left to right as: (Low, L) PTA for 500, 1000, and 2000 Hz; (Center, C) PTA for 3000, 4000 and 6000 Hz; and (High, H) HTL at 8000 Hz. This LCH Triad audiogram-classification system is evaluated using a large United States (U.S.) national dataset (N = 8,795) from adults 20 to 80 + years of age and two large clinical datasets totaling 8,254 adults covering a similar age range. Its ability to capture variations in hearing function was found to be superior to that of the widely used PTA4.

Descriptions of Hearing Loss Severity Differentially Influence Parental Concern about the Impact of Childhood Hearing Loss

PURPOSE

The purpose of this study was to measure how parent concern about childhood hearing loss varies under different description conditions: classification-based, audibility-based, and simulation-based descriptions.

METHOD

We randomly allocated study participants (n = 143) to complete an online survey about expected child difficulties with listening situations with hearing loss. Our participants were parents of children with typical hearing in the 0- to 12-month age range. Participants were exposed to one type of description (classification-based, audibility-based, or simulation-based) and one level of hearing loss (slight, mild, and moderate or their audibility and simulation equivalents), producing nine total groups. Participants rated the level of expected difficulty their child would experience performing age-appropriate listening tasks with the given hearing loss. They also selected what they perceived as the most appropriate intervention from a list of increasingly intense options.

RESULTS

Our findings revealed that audibility-based descriptions elicited significantly higher levels of parent concerns about hearing loss than classification-based strategies, but that simulation-based descriptions elicited the highest levels of concern. Those assigned to simulation-based and audibility-based groups also judged relatively more intense intervention options as appropriate compared to those assigned to classification-based groups.

CONCLUSIONS

This study expands our knowledge base about descriptive factors that impact levels of parent concern about hearing loss after diagnosis. This has potentially cascading effects on later intervention actions such as fitting hearing technology. It also provides a foundation for developing and testing clinical applications of audibility-based counseling strategies.

Data-driven audiogram classifier using data normalization and multi-stage feature selection

Audiograms are used to show the hearing capability of a person at different frequencies. The filter bank in a hearing aid is designed to match the shape of patients' audiograms. Configuring the hearing aid is done by modifying the designed filters' gains to match the patient's audiogram. There are few problems faced in achieving this objective successfully. There is a shortage in the number of audiologists; the filter bank hearing aid designs are complex; and, the hearing aid fitting process is tiring. In this work, a machine learning solution is introduced to classify the audiograms according to the shapes based on unsupervised spectral clustering. The features used to build the ML model are peculiar and describe the audiograms better. Different normalization methods are applied and studied statistically to improve the training data set. The proposed Machine Learning (ML) algorithm outperformed the current existing models, where, the accuracy, precision, recall, specificity, and F-score values are higher. The reason for the better performance is the use of multi-stage feature selection to describe the audiograms precisely. This work introduces a novel ML technique to classify audiograms according to the shape, which, can be integrated to the future and existing studies to change the existing practices in classifying audiograms.

Estimation of all-cause noise exposure for U.S. adults from national survey data

Millions of adults are at risk of hearing loss resulting from exposure to occupational and recreational noises. Data from the combined National Health and Nutrition Examination Survey (NHANES) 2011-2012 and 2015-2016 datasets were used to establish the prevalence of occupational and recreational noise exposures through self-report questions. For recreational noise exposures, NHANES asked about the use of firearms, including the use of hearing protection devices (HPDs) while shooting, and off-work exposures to very loud noise. For work exposures, NHANES asked about exposures to loud and very loud noise. For four of these five questions, graded responses on a 5- or 7-point scale were available. Receiver-operating-characteristic analyses were used to optimize the criterion response for identification of hearing loss for each question with graded responses using the unweighted data. Correlations among the graded responses supported reduction to two measures: (1) rounds fired combined with use of HPDs while shooting and (2) work exposure to loud and very loud noise combined. Logistic-regression analyses of various measures of pure-tone hearing loss were performed to examine the effects of recreational and occupational noise exposures on hearing loss. The odds of hearing loss were significantly greater for those who reported recreational and combined noise exposures.

Immediate Effects of (Simulated) Age-Related Hearing Loss on Cognitive Processing and Performance for the Backward-Digit-Span Task

The recall of auditorily presented sequences of digits in reverse order (also known as the Backward Digit Span, BDS) is considered to reflect a person's information storage and processing abilities which have been linked to speech-in-noise intelligibility. However, especially in aging research and audiology, persons who are administered the BDS task are often affected by hearing loss (HL). If uncorrected, HL can have immediate assessment-format-related effects on cognitive-test performance and can result, in the long term, in neuroplastic changes impacting cognitive functioning. In the present study, an impairment-simulation approach, mimicking mild-to-moderate age-related HLs typical for persons aged 65, 75, and 85 years, was used in 19 young normal-hearing participants to evaluate the impact of HL on cognitive performance and the cognitive processes probed by the BDS task. Participants completed the BDS task in several listening conditions, as well as several commonly used visual tests of short-term and working memory. The results indicated that BDS performance was impaired by a simulated HL representing that of persons aged 75 years and above. In the normal-hearing condition, BDS performance correlated positively with both performance on tests of short-term memory and performance on tests of working memory. In the listening condition simulating moderate HL (as experienced by the average 85-year-old person), BDS performance only correlated with performance on working-memory tests. In conclusion, simulated (and, by extrapolation, actual) age-related HL negatively affects cognitive-test performance and may change the composition of the cognitive processes associated with the completion of a cognitive task.

Using Automatic Speech Recognition to Optimize Hearing-Aid Time Constants

Automatic speech recognition (ASR), when combined with hearing-aid (HA) and hearing-loss (HL) simulations, can predict aided speech-identification performances of persons with age-related hearing loss. ASR can thus be used to evaluate different HA configurations, such as combinations of insertion-gain functions and compression thresholds, in order to optimize HA fitting for a given person. The present study investigated whether, after fixing compression thresholds and insertion gains, a random-search algorithm could be used to optimize time constants (i.e., attack and release times) for 12 audiometric profiles. The insertion gains were either those recommended by the CAM2 prescription rule or those optimized using ASR, while compression thresholds were always optimized using ASR. For each audiometric profile, the random-search algorithm was used to vary time constants with the aim to maximize ASR performance. A HA simulator and a HL simulator simulator were used, respectively, to amplify and to degrade speech stimuli according to the input audiogram. The resulting speech signals were fed to an ASR system for recognition. For each audiogram, 1,000 iterations of the random-search algorithm were used to find the time-constant configuration yielding the highest ASR score. To assess the reproducibility of the results, the random search algorithm was run twice. Optimizing the time constants significantly improved the ASR scores when CAM2 insertion gains were used, but not when using ASR-based gains. Repeating the random search yielded similar ASR scores, but different time-constant configurations.

OPRA-RS: A Hearing-Aid Fitting Method Based on Automatic Speech Recognition and Random Search

Hearing-aid (HA) prescription rules (such as NAL-NL2, DSL-v5, and CAM2) are used by HA audiologists to define initial HA settings (e.g., insertion gains, IGs) for patients. This initial fitting is later individually adjusted for each patient to improve clinical outcomes in terms of speech intelligibility and listening comfort. During this fine-tuning stage, speech-intelligibility tests are often carried out with the patient to assess the benefits associated with different HA settings. As these tests tend to be time-consuming and performance on them depends on the patient's level of fatigue and familiarity with the test material, only a limited number of HA settings can be explored. Consequently, it is likely that a suboptimal fitting is used for the patient. Recent studies have shown that automatic speech recognition (ASR) can be used to predict the effects of IGs on speech intelligibility for patients with age-related hearing loss (ARHL). The aim of the present study was to extend this approach by optimizing, in addition to IGs, compression thresholds (CTs). However, increasing the number of parameters to be fitted increases exponentially the number of configurations to be assessed. To limit the number of HA settings to be tested, three random-search (RS) genetic algorithms were used. The resulting new HA fitting method, combining ASR and RS, is referred to as "objective prescription rule based on ASR and random search" (OPRA-RS). Optimal HA settings were computed for 12 audiograms, representing average and individual audiometric profiles typical for various levels of ARHL severity, and associated ASR performances were compared to those obtained with the settings recommended by CAM2. Each RS algorithm was run twice to assess its reliability. For all RS algorithms, ASR scores obtained with OPRA-RS were significantly higher than those associated with CAM2. Each RS algorithm converged on similar optimal HA settings across repetitions. However, significant differences were observed between RS algorithms in terms of maximum ASR performance and processing costs. These promising results open the way to the use of ASR and RS algorithms for the fine-tuning of HAs with potential speech-intelligibility benefits for the patient.

Using Machine Learning and the National Health and Nutrition Examination Survey to Classify Individuals With Hearing Loss

Even before the COVID-19 pandemic, there was mounting interest in remote testing solutions for audiology. The ultimate goal of such work was to improve access to hearing healthcare for individuals that might be unable or reluctant to seek audiological help in a clinic. In 2015, Diane Van Tasell patented a method for measuring an audiogram when the precise signal level was unknown (patent US 8,968,209 B2). In this method, the slope between pure-tone thresholds measured at 2 and 4 kHz is calculated and combined with questionnaire information in order to reconstruct the most likely audiograms from a database of options. An approach like the Van Tasell method is desirable because it is quick and feasible to do in a patient's home where exact stimulus levels are unknown. The goal of the present study was to use machine learning to assess the effectiveness of such audiogram-estimation methods. The National Health and Nutrition Examination Survey (NHANES), a database of audiologic and demographic information, was used to train and test several machine learning algorithms. Overall, 9,256 cases were analyzed. Audiometric data were classified using the Wisconsin Age-Related Hearing Impairment Classification Scale (WARHICS), a method that places hearing loss into one of eight categories. Of the algorithms tested, a random forest machine learning algorithm provided the best fit with only a few variables: the slope between 2 and 4 kHz; gender; age; military experience; and self-reported hearing ability. Using this method, 54.79% of the individuals were correctly classified, 34.40% were predicted to have a milder loss than measured, and 10.82% were predicted to have a more severe loss than measured. Although accuracy was low, it is unlikely audibility would be severely affected if classifications were used to apply gains. Based on audibility calculations, underamplification still provided sufficient gain to achieve ~95% correct (Speech Intelligibility Index ≥ 0.45) for sentence materials for 88% of individuals. Fewer than 1% of individuals were overamplified by 10 dB for any audiometric frequency. Given these results, this method presents a promising direction toward remote assessment; however, further refinement is needed before use in clinical fittings.

Further Evaluation and Application of the Wisconsin Age-Related Hearing Impairment Classification System

Purpose The primary purpose of this report is to further evaluate the Wisconsin Age-Related Hearing Impairment Classification System (WARHICS) scheme for the classification of audiograms obtained from older adults, ages 50-89 years. Method This was a retrospective analysis of data from 11,729 older adults included in two population, one community, and two clinical data sets. To aid in the interpretation of the audiogram classifications, the functional significance of audiogram categories was evaluated using Hearing Handicap Inventory-Screen scores available for 10,330 of these same individuals. Results A WARHICS level could be obtained successfully for 99.5% of the 11,729 audiograms included in these analyses. It was applied successfully to population, community, and clinical samples. WARHICS Level 8 audiograms, corresponding to a broad profound hearing impairment (pure-tone thresholds from 500 to 8000 Hz, all > 80 dB HL), were very rare (.1% prevalence). WARHICS subtypes, designed to identify those with high-frequency notched audiograms or flat audiograms, occurred only 3%-6% of the time, depending on the data set. When compared to the current hearing-impairment grading system of the World Health Organization (WHO), several WARHICS levels fell within a given WHO hearing-impairment grade suggesting a finer grained categorization by the WARHICS. The Hearing Handicap Inventory-Screen scores provided justification for the fine-grained categorization according to the WARHICS. It is also demonstrated that the WARHICS levels could be extended to the classification of audiograms from all adults, regardless of age. Finally, a WARHICS-WHO hybrid classification scheme is proposed for future evaluation. Conclusions Although the WARHICS was developed from a population data set, it was demonstrated here to be applicable to population, community, and clinical data sets for adults aged 50-89 years. The applicability of the WARHICS or a WARHICS-WHO hybrid to population data sets that included younger adults suggests that this classification system has the potential to be applied even more broadly than just to age-related hearing loss.