Contributions and limitations of using machine learning to predict noise-induced hearing loss

Machine Learning Models Can Predict Tinnitus and Noise-Induced Hearing Loss

OBJECTIVES

Despite the extensive use of machine learning (ML) models in health sciences for outcome prediction and condition classification, their application in differentiating various types of auditory disorders remains limited. This study aimed to address this gap by evaluating the efficacy of five ML models in distinguishing (a) individuals with tinnitus from those without tinnitus and (b) noise-induced hearing loss (NIHL) from age-related hearing loss (ARHL).

DESIGN

We used data from a cross-sectional study of the Canadian population, which included audiologic and demographic information from 928 adults aged 30 to 100 years, diagnosed with either ARHL or NIHL due to long-term occupational noise exposure. The ML models applied in this study were artificial neural networks (ANNs), K-nearest neighbors, logistic regression, random forest (RF), and support vector machines.

RESULTS

The study revealed that tinnitus prevalence was over twice as high in the NIHL group compared with the ARHL group, with a frequency of 27.85% versus 8.85% in constant tinnitus and 18.55% versus 10.86% in intermittent tinnitus. In pattern recognition, significantly greater hearing loss was found at medium- and high-band frequencies in NIHL versus ARHL. In both NIHL and ARHL, individuals with tinnitus showed better pure-tone sensitivity than those without tinnitus. Among the ML models, ANN achieved the highest overall accuracy (70%), precision (60%), and F1-score (87%) for predicting tinnitus, with an area under the curve of 0.71. RF outperformed other models in differentiating NIHL from ARHL, with the highest precision (79% for NIHL, 85% for ARHL), recall (85% for NIHL), F1-score (81% for NIHL), and area under the curve (0.90).

CONCLUSIONS

Our findings highlight the application of ML models, particularly ANN and RF, in advancing diagnostic precision for tinnitus and NIHL, potentially providing a framework for integrating ML techniques into clinical audiology for improved diagnostic precision. Future research is suggested to expand datasets to include diverse populations and integrate longitudinal data.

Construction of a risk prediction model for occupational noise-induced hearing loss using routine blood and biochemical indicators in Shenzhen, China: a predictive modelling study

OBJECTIVES

Occupational noise-induced hearing loss (ONIHL) represents a prevalent occupational health condition, traditionally necessitating multiple pure-tone audiometry assessments. We have developed and validated a machine learning model leveraging routine haematological and biochemical parameters, thereby offering novel insights into the risk prediction of ONIHL.

DESIGN, SETTING AND PARTICIPANTS

This study analysed data from 3297 noise-exposed workers in Shenzhen, including 160 ONIHL cases, with the data set divided into D1 (2868 samples, 107 ONIHL cases) and D2 (429 samples, 53 ONIHL cases). The inclusion criteria were formulated based on the GBZ49-2014 Diagnosis of Occupational Noise-Induced Hearing Loss. Model training was performed using D1, and model validation was conducted using D2. Routine blood and biochemical indicators were extracted from the case data, and a range of machine learning algorithms including extreme gradient boosting (XGBoost) were employed to construct predictive models. The model underwent refinement to identify the most representative variables, and decision curve analysis was conducted to evaluate the net benefit of the model across various threshold levels.

PRIMARY OUTCOME MEASURES

Model creation data set and validation data sets: ONIHL.

RESULTS

The prediction model, developed using XGBoost, demonstrated exceptional performance, achieving an area under the receiver operating characteristic curve (AUC) of 0.942, a sensitivity of 0.875 and a specificity of 0.936 on the validation data set. On the test data set, the model achieved an AUC of 0.990. After implementing feature selection, the model was refined to include only 16 features, while maintaining strong performance on a newly acquired independent data set, with an AUC of 0.872, a balanced accuracy of 0.798, a sensitivity of 0.755 and a specificity of 0.840. The analysis of feature importance revealed that serum albumin (ALB), platelet distribution width (PDW), coefficient of variation in red cell distribution width (RDW-CV), serum creatinine (Scr) and lymphocyte percentage (LYMPHP) are critical factors for risk stratification in patients with ONIHL.

CONCLUSION

The analysis of feature importance identified ALB, PDW, RDW-CV, Scr and LYMPHP as pivotal factors for risk stratification in patients with ONIHL. The machine learning model, using XGBoost, effectively distinguishes patients with ONIHLamong individuals exposed to noise, thereby facilitating early diagnosis and intervention.

Hearing loss prediction equation for Iranian truck drivers using neural network algorithm

BACKGROUND

 Given the high prevalence of hearing loss among truck drivers, using artificial neural networks (ANNs) to predict and detect contributing factors early can aid managers significantly.

OBJECTIVE

 This study aimed to predict hearing loss using an ANN algorithm and to evaluate the weight and influence of various factors affecting hearing loss among truck drivers.

METHODS

 A total of 692 truck drivers were selected for the study. Their occupational exposure histories were collected to identify factors influencing their hearing loss. The impact and weight of each factor were measured, and an ANN algorithm was used to model and predict the degree of hearing loss.

RESULTS

 The assessment of hearing loss among truck drivers revealed a prevalence of 59.98% in the right ear and 64.74% in the left ear. The most significant average hearing loss in both ears occurred at frequencies of 6000 and 8000 Hz. According to the ANN model, age and the frequency of 2000 Hz had the greatest impact on hearing loss, while sound pressure level (SPL) had the least impact. Additionally, the relationship between overall hearing loss and the type of heavy truck indicated that drivers of HOWO brand trucks experienced the highest degree of hearing loss compared to other drivers.

CONCLUSIONS

 This study demonstrates that the ANN algorithm is a promising tool for predicting hearing impairments caused by noise exposure among truck drivers.

Machine learning analysis of cardiovascular risk factors and their associations with hearing loss

Hearing loss poses immense burden worldwide and early detection is crucial. The accurate models identify high-risk groups, enabling timely intervention to improve quality of life. The subtle changes in hearing often go unnoticed, presenting a challenge for early hearing loss detection. While machine learning shows promise, prior studies have not leveraged cardiovascular risk factors known to impact hearing. As hearing outcomes remain challenging to characterize associations, we evaluated a new approach to predict current hearing outcomes through machine learning models using cardiovascular risk factors. The National Health and Nutrition Examination Survey (NHANES) 2012-2018 data comprising audiometric tests and cardiovascular risk factors was utilized. Machine learning algorithms were trained to classify hearing impairment thresholds and predict pure tone average values. Key results showed light gradient boosted machine performing best in classifying mild or greater impairment (> 25 dB HL) with 80.1% accuracy. It also classified > 16 dB HL and > 40 dB HL thresholds, with accuracies exceeding 77% and 86% respectively. The study also found that CatBoost and Gradient Boosting performed well in classifying hearing loss thresholds, with test set accuracies around 0.79 and F1-scores around 0.79-0.80. A multi-layer neural network emerged as the top predictor of pure tone averages, achieving a mean absolute error of just 3.05 dB. Feature analysis identified age, gender, blood pressure and waist circumference as key associated factors. Findings offer a promising direction for a clinically applicable tool, personalized prevention strategies, and calls for prospective validation.

Research on noise-induced hearing loss based on functional and structural MRI using machine learning methods

Noise-induced hearing loss (NIHL) is a common occupational condition. The aim of this study was to develop a classification model for NIHL on the basis of both functional magnetic resonance imaging (fMRI) and structural magnetic resonance imaging (sMRI) by applying machine learning methods. fMRI indices such as the amplitude of low-frequency fluctuation (ALFF), fractional amplitude of low-frequency fluctuation (fALFF), regional homogeneity (ReHo), degree of centrality (DC), and sMRI indices such as gray matter volume (GMV), white matter volume (WMV), and cortical thickness were extracted from each brain region. The least absolute shrinkage and selection operator was used to reduce and select the optimal features. The support vector machine (SVM), random forest (RF) and logistic regression (LR) algorithms, were used to establish the classification model for NIHL. Finally, the SVM model based on combined fMRI indices, achieved the best performance, with area under the receiver operating characteristic curve of 0.97 and an accuracy of 95%. The SVM classification model that integrates fMRI indicators has the greatest potential for identifying NIHL patients and healthy people, revealing the complementary role of fMRI indicators in classification and indicating that it is necessary to include multiple indicators of the brain when establishing a classification model.

Applications of Machine Learning in Meniere's Disease Assessment Based on Pure-Tone Audiometry

OBJECTIVE

To apply machine learning models based on air conduction thresholds of pure-tone audiometry for automatic diagnosis of Meniere's disease (MD) and prediction of endolymphatic hydrops (EH).

STUDY DESIGN

Retrospective study.

SETTING

Tertiary medical center.

METHODS

Gadolinium-enhanced magnetic resonance imaging sequences and pure-tone audiometry data were collected. Subsequently, basic and multiple analytical features were engineered based on the air conduction thresholds of pure-tone audiometry. Later, 5 classical machine learning models were trained to diagnose MD using the engineered features. The models demonstrating excellent performance were also selected to predict EH. The model's effectiveness in MD diagnosis was compared with experienced otolaryngologists.

RESULTS

First, the winning light gradient boosting (LGB) machine learning model trained by multiple features demonstrates a remarkable performance on the diagnosis of MD, achieving an accuracy rate of 87%, sensitivity of 83%, specificity of 90%, and a robust area under the receiver operating characteristic curve of 0.95, which compares favorably with experienced clinicians. Second, the LGB model, with an accuracy of 78% on EH prediction, outperformed the other 3 machine learning models. Finally, a feature importance analysis reveals a pivotal role of the specific pure-tone audiometry features that are essential for both MD diagnosis and EH prediction. Highlighted features include standard deviation and mean of the whole-frequency hearing, the peak of the audiogram, and hearing at low frequencies, notably at 250 Hz.

CONCLUSION

An efficient machine learning model based on pure-tone audiometry features was produced to diagnose MD, which also showed the potential to predict the subtypes of EH. The innovative approach demonstrated a game-changing strategy for MD screening and promising cost-effective benefits for the health care enterprise.

Evaluating Prediction Models with Hearing Handicap Inventory for the Elderly in Chronic Otitis Media Patients

BACKGROUND

This retrospective, cross-sectional study aimed to assess the functional hearing capacity of individuals with Chronic Otitis Media (COM) using prediction modeling techniques and the Hearing Handicap Inventory for the Elderly (HHIE) questionnaire. This study investigated the potential of predictive models to identify hearing levels in patients with COM.

METHODS

We comprehensively examined 289 individuals diagnosed with COM, of whom 136 reported tinnitus and 143 did not. This study involved a detailed analysis of various patient characteristics and HHIE questionnaire results. Logistic and Random Forest models were employed and compared based on key performance metrics.

RESULTS

The logistic model demonstrated a slightly higher accuracy (73.56%), area under the curve (AUC; 0.73), Kappa value (0.45), and F1 score (0.78) than the Random Forest model. These findings suggest the superior predictive performance of the logistic model in identifying hearing levels in patients with COM.

CONCLUSIONS

Although the AUC for the logistic regression did not meet the benchmark, this study highlights the potential for enhanced reliability and improved performance metrics using a larger dataset. The integration of prediction modeling techniques and the HHIE questionnaire shows promise for achieving greater diagnostic accuracy and refining intervention strategies for individuals with COM.

Combination of static and dynamic neural imaging features to distinguish sensorineural hearing loss: a machine learning study

Purpose

Sensorineural hearing loss (SNHL) is the most common form of sensory deprivation and is often unrecognized by patients, inducing not only auditory but also nonauditory symptoms. Data-driven classifier modeling with the combination of neural static and dynamic imaging features could be effectively used to classify SNHL individuals and healthy controls (HCs).

Methods

We conducted hearing evaluation, neurological scale tests and resting-state MRI on 110 SNHL patients and 106 HCs. A total of 1,267 static and dynamic imaging characteristics were extracted from MRI data, and three methods of feature selection were computed, including the Spearman rank correlation test, least absolute shrinkage and selection operator (LASSO) and t test as well as LASSO. Linear, polynomial, radial basis functional kernel (RBF) and sigmoid support vector machine (SVM) models were chosen as the classifiers with fivefold cross-validation. The receiver operating characteristic curve, area under the curve (AUC), sensitivity, specificity and accuracy were calculated for each model.

Results

SNHL subjects had higher hearing thresholds in each frequency, as well as worse performance in cognitive and emotional evaluations, than HCs. After comparison, the selected brain regions using LASSO based on static and dynamic features were consistent with the between-group analysis, including auditory and nonauditory areas. The subsequent AUCs of the four SVM models (linear, polynomial, RBF and sigmoid) were as follows: 0.8075, 0.7340, 0.8462 and 0.8562. The RBF and sigmoid SVM had relatively higher accuracy, sensitivity and specificity.

Conclusion

Our research raised attention to static and dynamic alterations underlying hearing deprivation. Machine learning-based models may provide several useful biomarkers for the classification and diagnosis of SNHL.

Machine learning-based longitudinal prediction for GJB2-related sensorineural hearing loss

BACKGROUND

Recessive GJB2 variants, the most common genetic cause of hearing loss, may contribute to progressive sensorineural hearing loss (SNHL). The aim of this study is to build a realistic predictive model for GJB2-related SNHL using machine learning to enable personalized medical planning for timely intervention.

METHOD

Patients with SNHL with confirmed biallelic GJB2 variants in a nationwide cohort between 2005 and 2022 were included. Different data preprocessing protocols and computational algorithms were combined to construct a prediction model. We randomly divided the dataset into training, validation, and test sets at a ratio of 72:8:20, and repeated this process ten times to obtain an average result. The performance of the models was evaluated using the mean absolute error (MAE), which refers to the discrepancy between the predicted and actual hearing thresholds.

RESULTS

We enrolled 449 patients with 2184 audiograms available for deep learning analysis. SNHL progression was identified in all models and was independent of age, sex, and genotype. The average hearing progression rate was 0.61 dB HL per year. The best MAE for linear regression, multilayer perceptron, long short-term memory, and attention model were 4.42, 4.38, 4.34, and 4.76 dB HL, respectively. The long short-term memory model performed best with an average MAE of 4.34 dB HL and acceptable accuracy for up to 4 years.

CONCLUSIONS

We have developed a prognostic model that uses machine learning to approximate realistic hearing progression in GJB2-related SNHL, allowing for the design of individualized medical plans, such as recommending the optimal follow-up interval for this population.

A machine learning screening model for identifying the risk of high-frequency hearing impairment in a general population

BACKGROUND

Hearing impairment (HI) has become a major public health issue in China. Currently, due to the limitations of primary health care, the gold standard for HI diagnosis (pure-tone hearing test) is not suitable for large-scale use in community settings. Therefore, the purpose of this study was to develop a cost-effective HI screening model for the general population using machine learning (ML) methods and data gathered from community-based scenarios, aiming to help improve the hearing-related health outcomes of community residents.

METHODS

This study recruited 3371 community residents from 7 health centres in Zhejiang, China. Sixty-eight indicators derived from questionnaire surveys and routine haematological tests were delivered and used for modelling. Seven commonly used ML models (the naive Bayes (NB), K-nearest neighbours (KNN), support vector machine (SVM), random forest (RF), eXtreme Gradient Boosting (XGBoost), boosting, and least absolute shrinkage and selection operator (LASSO regression)) were adopted and compared to develop the final high-frequency hearing impairment (HFHI) screening model for community residents. The model was constructed with a nomogram to obtain the risk score of the probability of individuals suffering from HFHI. According to the risk score, the population was divided into three risk stratifications (low, medium and high) and the risk factor characteristics of each dimension under different risk stratifications were identified.

RESULTS

Among all the algorithms used, the LASSO-based model achieved the best performance on the validation set by attaining an area under the curve (AUC) of 0.868 (95% confidence interval (CI): 0.847-0.889) and reaching precision, specificity and F-score values all greater than 80%. Five demographic indicators, 7 disease-related features, 5 behavioural factors, 2 environmental exposures, 2 hearing cognitive factors, and 13 blood test indicators were identified in the final screening model. A total of 91.42% (1235/1129) of the subjects in the high-risk group were confirmed to have HI by audiometry, which was 3.99 times greater than that in the low-risk group (22.91%, 301/1314). The high-risk population was mainly characterized as older, low-income and low-educated males, especially those with multiple chronic conditions, noise exposure, poor lifestyle, abnormal blood indices (e.g., red cell distribution width (RDW) and platelet distribution width (PDW)) and liver function indicators (e.g., triglyceride (TG), indirect bilirubin (IBIL), aspartate aminotransferase (AST) and low-density lipoprotein (LDL)). An HFHI nomogram was further generated to improve the operability of the screening model for community applications.

CONCLUSIONS

The HFHI risk screening model developed based on ML algorithms can more accurately identify residents with HFHI by categorizing them into the high-risk groups, which can further help to identify modifiable and immutable risk factors for residents at high risk of HI and promote their personalized HI prevention or intervention.

Quantum computing and machine learning for Arabic language sentiment classification in social media

With the increasing amount of digital data generated by Arabic speakers, the need for effective and efficient document classification techniques is more important than ever. In recent years, both quantum computing and machine learning have shown great promise in the field of document classification. However, there is a lack of research investigating the performance of these techniques on the Arabic language. This paper presents a comparative study of quantum computing and machine learning for two datasets of Arabic language document classification. In the first dataset of 213,465 Arabic tweets, both classic machine learning (ML) and quantum computing approaches achieve high accuracy in sentiment analysis, with quantum computing slightly outperforming classic ML. Quantum computing completes the task in approximately 59 min, slightly faster than classic ML, which takes around 1 h. The precision, recall, and F1 score metrics indicate the effectiveness of both approaches in predicting sentiment in Arabic tweets. Classic ML achieves precision, recall, and F1 score values of 0.8215, 0.8175, and 0.8121, respectively, while quantum computing achieves values of 0.8239, 0.8199, and 0.8147, respectively. In the second dataset of 44,000 tweets, both classic ML (using the Random Forest algorithm) and quantum computing demonstrate significantly reduced processing times compared to the first dataset, with no substantial difference between them. Classic ML completes the analysis in approximately 2 min, while quantum computing takes approximately 1 min and 53 s. The accuracy of classic ML is higher at 0.9241 compared to 0.9205 for quantum computing. However, both approaches achieve high precision, recall, and F1 scores, indicating their effectiveness in accurately predicting sentiment in the dataset. Classic ML achieves precision, recall, and F1 score values of 0.9286, 0.9241, and 0.9249, respectively, while quantum computing achieves values of 0.92456, 0.9205, and 0.9214, respectively. The analysis of the metrics indicates that quantum computing approaches are effective in identifying positive instances and capturing relevant sentiment information in large datasets. On the other hand, traditional machine learning techniques exhibit faster processing times when dealing with smaller dataset sizes. This study provides valuable insights into the strengths and limitations of quantum computing and machine learning for Arabic document classification, emphasizing the potential of quantum computing in achieving high accuracy, particularly in scenarios where traditional machine learning techniques may encounter difficulties. These findings contribute to the development of more accurate and efficient document classification systems for Arabic data.

Ensemble filters with harmonize PSO-SVM algorithm for optimal hearing disorder prediction

Discovering a hearing disorder at an earlier intervention is critical for reducing the effects of hearing loss and the approaches to increase the remaining hearing ability can be implemented to achieve the successful development of human communication. Recently, the explosive dataset features have increased the complexity for audiologists to decide the proper treatment for the patient. In most cases, data with irrelevant features and improper classifier parameters causes a crucial influence on the audiometry system in terms of accuracy. This is due to the dependent processes of these two, where the classification accuracy performance could be worsened if both processes are conducted independently. Although the filter algorithm is capable of eliminating irrelevant features, it still lacks the ability to consider feature reliance and results in a poor selection of significant features. Improper kernel parameter settings may also contribute to poor accuracy performance. In this paper, an ensemble filters feature selection based on Information Gain (IG), Gain Ratio (GR), Chi-squared (CS), and Relief-F (RF) with harmonize optimization of Particle Swarm Optimization (PSO) and Support Vector Machine (SVM) is presented to mitigate these problems. Ensemble filters are utilized so that the initial top dominant features relevant for classification can be considered. Then, PSO and SVM are optimized simultaneously to achieve the optimal solution. The results on a standard Audiology dataset show that the proposed method produces 96.50% accuracy with optimal solution compared to classical SVM, which signifies the proposed method is effective in handling high dimensional data for hearing disorder prediction.

Machine Learning for the Orthopaedic Surgeon: Uses and Limitations

➤

Machine learning is a subset of artificial intelligence in which computer algorithms are trained to make classifications and predictions based on patterns in data. The utilization of these techniques is rapidly expanding in the field of orthopaedic research.

➤

There are several domains in which machine learning has application to orthopaedics, including radiographic diagnosis, gait analysis, implant identification, and patient outcome prediction.

➤

Several limitations prevent the widespread use of machine learning in the daily clinical environment. However, future work can overcome these issues and enable machine learning tools to be a useful adjunct for orthopaedic surgeons in their clinical decision-making.

A Deep Learning Approach to Predict Conductive Hearing Loss in Patients With Otitis Media With Effusion Using Otoscopic Images

Importance

Otitis media with effusion (OME) is one of the most common causes of acquired conductive hearing loss (CHL). Persistent hearing loss is associated with poor childhood speech and language development and other adverse consequence. However, to obtain accurate and reliable hearing thresholds largely requires a high degree of cooperation from the patients.

Objective

To predict CHL from otoscopic images using deep learning (DL) techniques and a logistic regression model based on tympanic membrane features.

Design, Setting, and Participants

A retrospective diagnostic/prognostic study was conducted using 2790 otoscopic images obtained from multiple centers between January 2015 and November 2020. Participants were aged between 4 and 89 years. Of 1239 participants, there were 209 ears from children and adolescents (aged 4-18 years [16.87%]), 804 ears from adults (aged 18-60 years [64.89%]), and 226 ears from older people (aged >60 years, [18.24%]). Overall, 679 ears (54.8%) were from men. The 2790 otoscopic images were randomly assigned into a training set (2232 [80%]), and validation set (558 [20%]). The DL model was developed to predict an average air-bone gap greater than 10 dB. A logistic regression model was also developed based on otoscopic features.

Main Outcomes and Measures

The performance of the DL model in predicting CHL was measured using the area under the receiver operating curve (AUC), accuracy, and F1 score (a measure of the quality of a classifier, which is the harmonic mean of precision and recall; a higher F1 score means better performance). In addition, these evaluation parameters were compared to results obtained from the logistic regression model and predictions made by three otologists.

Results

The performance of the DL model in predicting CHL showed the AUC of 0.74, accuracy of 81%, and F1 score of 0.89. This was better than the results from the logistic regression model (ie, AUC of 0.60, accuracy of 76%, and F1 score of 0.82), and much improved on the performance of the 3 otologists; accuracy of 16%, 30%, 39%, and F1 scores of 0.09, 0.18, and 0.25, respectively. Furthermore, the DL model took 2.5 seconds to predict from 205 otoscopic images, whereas the 3 otologists spent 633 seconds, 645 seconds, and 692 seconds, respectively.

Conclusions and Relevance

The model in this diagnostic/prognostic study provided greater accuracy in prediction of CHL in ears with OME than those obtained from the logistic regression model and otologists. This indicates great potential for the use of artificial intelligence tools to facilitate CHL evaluation when CHL is unable to be measured.