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.

Download Full-text

Age-related hearing loss in the Korea National Health and Nutrition Examination Survey

PLoS ONE ◽

10.1371/journal.pone.0243001 ◽

2020 ◽

Vol 15 (12) ◽

pp. e0243001

Author(s):

Subin Kim ◽

Jung Mee Park ◽

Jae Sang Han ◽

Jae Hyun Seo ◽

Kyung-Do Han ◽

...

Keyword(s):

Hearing Loss ◽

Hearing Impairment ◽

National Health ◽

High Frequency ◽

Korean Population ◽

Nutrition Examination Survey ◽

Health And Nutrition ◽

Age Related ◽

Age Related Hearing Loss ◽

Underlying Diseases

Objectives Age-related hearing loss (ARHL), also known as presbycusis, is a chronic disorder characterized by impairment of the transduction of acoustic signals. This study analysed the prevalence and demographic characteristics of ARHL in the Korean population. Methods We used the data from the Korea National Health and Nutrition Examination Survey (KNHANES) from 2009 to 2012 and analysed the association between age and hearing impairment. A total of 16,799 adults were selected for the current study. Physical examinations, blood tests, otoscopic examinations, and hearing tests were performed. The demographic variables included age, gender, obesity, economic status, education level, noise exposure history, and underlying diseases. Results Among 16,799 participants, the prevalence of unilateral hearing loss was 8% (1,349 people), and bilateral hearing loss was 5.9% (989 people). Men were 53.4% more likely to have hearing loss than women. Age and underlying diseases, like hypertension, diabetes, and abdominal obesity, were significantly associated with hearing loss (P < 0.0001). Further, mental health factors, such as cognitive function, depression, and suicidal ideation, were related to hearing loss. The prevalence of hearing loss increased with advancing years, especially in the high frequency of 6 kHz, with a sharply increase in patients aged 65 and over. Conclusion The analysis of auditory performance in the Korean population confirmed the association of high-frequency hearing loss with advancing age. A threshold of 6 kHz should be included to correctly diagnose hearing impairment in elderly patients. Patients with ARHL should be provided with suitable aural rehabilitation that includes active high-frequency control.

Download Full-text

Associations of Dietary Riboflavin, Niacin, and Retinol with Age-related Hearing Loss: An Analysis of Korean National Health and Nutrition Examination Survey Data

Nutrients ◽

10.3390/nu11040896 ◽

2019 ◽

Vol 11 (4) ◽

pp. 896 ◽

Cited By ~ 3

Author(s):

Tae Su Kim ◽

Jong Woo Chung

Keyword(s):

Hearing Loss ◽

National Health ◽

Smoking Status ◽

The Elderly ◽

Nutrition Examination Survey ◽

Modifying Factors ◽

Health And Nutrition ◽

Age Related ◽

Using Data ◽

Age Related Hearing Loss

Because age-related hearing loss (ARHL) is irreversible, prevention is very important. Thus, investigating modifying factors that help prevent ARHL is critical for the elderly. Nutritional status or nutritional factors for the elderly are known to be associated with many problems related to aging. Emerging studies suggest that there was the interaction between nutrition and ARHL. We aimed to investigate the possible impact of dietary nutrients on ARHL using data from the fifth Korean National Health and Nutrition Examination Survey (KNHANES) which included 4742 subjects aged ≥ 65 years from 2010 to 2012. All participants underwent an otologic examination, audiologic evaluation, and nutritional survey. The associations between ARHL and nutrient intake were analyzed using simple and multiple regression models with complex sampling adjusted for confounding factors, such as BMI, smoking status, alcohol consumption, and history of hypertension and diabetes. Higher intake groups of riboflavin, niacin and retinol was inversely associated with ARHL prevalence (riboflavin aOR, 0.71; 95% CI, 0.54–0.94; p = 0.016, niacin aOR, 0.72; 95% CI, 0.54–0.96; p = 0.025, retinol aOR 0.66; 95% CI, 0.51–0.86; p = 0.002, respectively). Our findings suggest the recommended intake levels of riboflavin, niacin, and retinol may help reduce ARHL in the elderly.

Download Full-text

Cross-sectional study of the association between age-related macular degeneration and arthritis in the National Health and Nutrition Examination Survey 2005–2008

BMJ Open ◽

10.1136/bmjopen-2019-035805 ◽

2020 ◽

Vol 10 (12) ◽

pp. e035805

Author(s):

Zhuoting Zhu ◽

Huan Liao ◽

Sen Liu ◽

Jian Zhang ◽

Yifan Chen ◽

...

Keyword(s):

Macular Degeneration ◽

National Health ◽

Logistic Models ◽

Cross Sectional Study ◽

Age Related Macular Degeneration ◽

Nutrition Examination Survey ◽

Sectional Study ◽

Cross Sectional ◽

Health And Nutrition ◽

Age Related

ObjectiveTo explore the association between age-related macular degeneration (AMD) and arthritis in a representative sample of the US population.DesignPopulation-based, cross-sectional study.SettingThe National Health and Nutrition Examination Survey (NHANES) 2005–2008.ParticipantsA total of 4813 participants aged 40 years and older with available information on AMD and arthritis in the 2005–2008 NHANES.MethodsThe status and types of arthritis were obtained from questionnaires. Non-mydriatic fundus photographs were collected. The types of AMD were assessed using the modified Wisconsin Age-Related Maculopathy Grading Classification Scheme. The association between arthritis and AMD was evaluated using logistic regression models.ResultsAfter adjusting for covariates, participants with any or early AMD had significantly lower odds of having any type of arthritis (any AMD: OR=0.56, 95% CI: 0.36–0.86; early AMD: OR=0.55, 95% CI: 0.34–0.88) or osteoarthritis (OA) (any AMD: OR=0.43, 95% CI: 0.26–0.71; early AMD: OR=0.44, 95% CI: 0.25–0.76) compared with those without AMD. When considering AMD as the outcome, significant negative associations were also found between any arthritis or OA and any (any arthritis: OR=0.64, 95% CI: 0.43–0.94; OA: OR=0.52, 95% CI: 0.33–0.82) or early AMD (any arthritis: OR=0.61, 95% CI: 0.40–0.93; OA: OR=0.51, 95% CI: 0.31–0.86) in the multivariable logistic models. There was no significant association between different types of arthritis and late AMD.ConclusionsPeople with arthritis, especially those with OA, were less likely to have AMD compared with those without arthritis and vice versa. Further studies are needed to confirm this potential protective effect of arthritis and/or arthritis treatment on AMD and to explore the underlying mechanisms.

Download Full-text

Age-related Smell and Taste Impairments and Vitamin D Associations in the U.S. Adults National Health and Nutrition Examination Survey

Nutrients ◽

10.3390/nu12040984 ◽

2020 ◽

Vol 12 (4) ◽

pp. 984

Author(s):

Galya Bigman

Keyword(s):

Vitamin D ◽

National Health ◽

The Elderly ◽

Cross Sectional Study ◽

Nutritional Deficiencies ◽

Nutrition Examination Survey ◽

Cross Sectional ◽

Health And Nutrition ◽

Age Related ◽

The Us

Smell and taste decline with aging, and markedly deteriorate when nutritional deficiencies occur. This study aims to examine the associations between Vitamin D (VD) deficiency and smell and taste impairments among adults. This paper details a cross-sectional study utilizing data from the US National Health and Nutrition Examination Survey (NHANES, 2013–2014.). Smell impairment was assessed by the Pocket Smell Test and defined as failing to correctly identify six or more of the eight odors. Taste impairment was defined as failing to correctly identify quinine or sodium chloride. VD was measured as serum 25-hydroxyvitamin. Multivariable weighted logistic regressions were utilized. Adjusted odds ratio (OR) and 95% confidence interval (CI) were presented. Overall, 2216 (smell sample) and 2636 (taste sample) participants were included, aged between 40 and 80 years old. Of those, 18.3% had taste impairment, 12.2% had smell impairment, and 20% had VD deficiency (<20 ng/mL). Compared to participants with sufficient VD (>30 ng/mL), those with VD deficiency were more likely by 39% to report a higher prevalence of smell impairment (OR = 1.39, 95%CI: 1.02–1.89); and only participants aged 70–80 years with VD inadequacy (20–30 ng/mL) were more likely by 96% to report a higher prevalence of taste impairment (OR = 1.96, 95%CI: 1.35–1.85). VD may have a significant role in age-related smell impairment in adults aged 40 years or older, and in age-related taste impairment in the elderly aged 70–80 years.

Download Full-text

A combined strategy of feature selection and machine learning to identify predictors of prediabetes

Journal of the American Medical Informatics Association ◽

10.1093/jamia/ocz204 ◽

2019 ◽

Vol 27 (3) ◽

pp. 396-406 ◽

Cited By ~ 1

Author(s):

Kushan De Silva ◽

Daniel Jönsson ◽

Ryan T Demmer

Keyword(s):

Machine Learning ◽

Feature Selection ◽

National Health ◽

Screening Tool ◽

Model Performance ◽

Nutrition Examination Survey ◽

Validation Data ◽

Internal Validation ◽

Health And Nutrition ◽

Wide Range

Abstract Objective To identify predictors of prediabetes using feature selection and machine learning on a nationally representative sample of the US population. Materials and Methods We analyzed n = 6346 men and women enrolled in the National Health and Nutrition Examination Survey 2013–2014. Prediabetes was defined using American Diabetes Association guidelines. The sample was randomly partitioned to training (n = 3174) and internal validation (n = 3172) sets. Feature selection algorithms were run on training data containing 156 preselected exposure variables. Four machine learning algorithms were applied on 46 exposure variables in original and resampled training datasets built using 4 resampling methods. Predictive models were tested on internal validation data (n = 3172) and external validation data (n = 3000) prepared from National Health and Nutrition Examination Survey 2011–2012. Model performance was evaluated using area under the receiver operating characteristic curve (AUROC). Predictors were assessed by odds ratios in logistic models and variable importance in others. The Centers for Disease Control (CDC) prediabetes screening tool was the benchmark to compare model performance. Results Prediabetes prevalence was 23.43%. The CDC prediabetes screening tool produced 64.40% AUROC. Seven optimal (≥ 70% AUROC) models identified 25 predictors including 4 potentially novel associations; 20 by both logistic and other nonlinear/ensemble models and 5 solely by the latter. All optimal models outperformed the CDC prediabetes screening tool (P < 0.05). Discussion Combined use of feature selection and machine learning increased predictive performance outperforming the recommended screening tool. A range of predictors of prediabetes was identified. Conclusion This work demonstrated the value of combining feature selection with machine learning to identify a wide range of predictors that could enhance prediabetes prediction and clinical decision-making.

Download Full-text