Related Experiment Video
Updated: Jul 8, 2025
07:13
Modified Experimental Conditions for Noise-Induced Hearing Loss in Mice and Assessment of Hearing Function and Outer Hair Cell Damage
Published on: February 10, 2023
2.3K
Optimizing age-related hearing risk predictions: an advanced machine learning integration with HHIE-S
Tzong-Hann Yang1,2,3,4, Yu-Fu Chen3, Yen-Fu Cheng5,6,7,8
1Department of Otorhinolaryngology, Taipei City Hospital, Taipei, 100, Taiwan.
Biodata Mining
|December 15, 2023
Summary
Machine learning enhances the accuracy of the Hearing Handicap Inventory for the Elderly Screening (HHIE-S) tool for detecting age-related hearing loss (ARHL) in older adults. This novel approach improves early detection and intervention for ARHL.
Area of Science:
- Gerontology and Audiology
- Computational Health Sciences
- Biomedical Data Science
Background:
- Age-related hearing loss (ARHL) disproportionately affects the elderly population.
- The Hearing Handicap Inventory for the Elderly Screening (HHIE-S) is a traditional tool for ARHL evaluation, primarily used for direct screening via self-reported outcomes.
- Existing methods for ARHL assessment often lack the predictive accuracy needed for timely intervention in older adults.
Purpose of the Study:
- To investigate the novel integration of machine learning (ML) approaches with the HHIE-S tool.
- To improve the predicted accuracy of the HHIE-S for identifying ARHL in the elderly population.
- To leverage advanced ML models for more precise ARHL assessment in senior citizens.
Main Methods:
- Utilized a dataset of 1,526 senior citizens from Taipei City Hospital (2016-2018).
- Employed machine learning models including XGBoost, Gradient Boosting, and LightGBM (LGBM) on training (n=1220) and testing (n=356) data subsets.
- Selected the LGBM model due to its superior Area Under the Curve (AUC) of 0.83, employing SHAP values for enhanced model interpretability and to prevent overfitting.
Main Results:
- The LGBM model achieved a strong AUC of 0.82 on the testing set, significantly outperforming conventional methods.
- Key significant predictors identified included specific HHIE-S items and participant age.
- This study uniquely combines ML (LGBM classifier) with the HHIE-S, moving beyond traditional focus on psychological effects of hearing loss.
Conclusions:
- Integrating machine learning with validated hearing evaluation instruments like HHIE-S shows significant potential for improved ARHL assessment.
- This ML-enhanced approach offers healthcare practitioners more accurate ARHL prediction capabilities.
- Facilitates earlier and more precise interventions for age-related hearing loss in the elderly.
