.png){kind=icon}
.png){kind=icon}
Return
HDDF: an ensemble learning framework for accurate hearing disorder detection using optimized feature selection
.png){kind=icon}
PRE
AI
S
F
S
DOI：10.1007/s10586-026-06457-3.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Hearing is crucial for daily communication and overall well-being. Hearing loss is a serious issue for individuals of all ages and ethnicities. Early detection of a hearing disorder is crucial for minimizing the impact of hearing loss and enhancing communication skills. This study introduces a novel Framework to discover hearing disorders known as the Hearing Disorder Discovering Framework (HDDF). The suggested HDDF comprises two stages: (i) preparation and (ii) discovery. During the preparation stage, the best beneficial features are chosen using the Crossover-based Binary Salp Swarm Algorithm (CBSSA). In contrast, in the discovery stage, ensemble classification is utilized to discover novel instances by combining results from three different classifiers: (a) Long Short-Term Memory (LSTM), (b) Ranked K-Nearest Neighbors (RKNN), and (c) Support Vector Machine (SVM). Furthermore, the proposed classifiers’ recommendations are integrated into an advanced voting mechanism known as the fuzzy-based judgment procedure. HDDF has been evaluated against modern discovery methodologies. Experimental results show that HDDF beats other hearing discovery methodologies by introducing the best precise assessment according to the audiology dataset. Experimental results show that HDDF outperforms existing methods, achieving the highest accuracy of 94.3% and an AUC-ROC of 0.97. These findings demonstrate the robustness and reliability of HDDF in detecting hearing disorders with high precision.
Keywords:
Hearing disorder detection
Ensemble learning
Feature selection
Crossover-based Binary Salp Swarm Algorithm (CBSSA)
Journal
C
IF:
4.1
Papers:
4.8K
Citations:
7.5K
Organization
Cited Papers