Utilizing Artificial Neural Networks for Establishing Hearing-Loss Predicting Models Based on a Longitudinal Dataset...

Area of Science:

• Occupational Health
• Audiology
• Data Science

Background:

• Traditional artificial neural network (ANN) models for noise-induced hearing loss (NIHL) often use cross-sectional data, limiting their ability to capture the chronic nature of hearing damage from long-term occupational noise exposure.
• This limitation necessitates the development of more robust predictive models that account for the temporal progression of NIHL.

Purpose of the Study:

• To develop and evaluate artificial neural network (ANN) models for predicting noise-induced hearing loss (NIHL) using a comprehensive eight-year longitudinal dataset.
• To assess the advantages of using longitudinal data and a more extensive variable set for improving NIHL prediction accuracy in industrial workers.

Main Methods:

• Utilized an eight-year longitudinal dataset of personal hearing threshold levels (HTLs) from 170 steel industry workers.
• Incorporated personal (7) and environmental (2) variables, including noise patterns and personal protective equipment (PPE) usage, into ANN models.
• Compared prediction performance using longitudinal data versus sub-datasets derived from cross-sectional data.

Main Results:

• Longitudinal datasets demonstrated superior predictive capabilities for NIHL compared to cross-sectional datasets.
• Inclusion of a comprehensive dataset, considering factors like noise patterns and PPE use, significantly enhanced NIHL prediction accuracy.
• ANN models consistently identified a discernible trend in NIHL progression, despite individual fluctuations in measured HTLs.

Conclusions:

• The developed ANN models show a consistent correlation with measured hearing threshold levels (HTLs), offering a valuable tool for industry.
• While caution is advised for individual worker predictions due to HTL variability, these models provide a reliable reference for managing industrial hearing conservation programs.
• Longitudinal data analysis with ANN techniques represents a significant advancement in predicting and managing occupational noise-induced hearing loss.