Hearing Device Fitting System Using Deep Neural Network Ensemble
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hearing device fitting methods, especially for over-the-counter devices, lack individualization and flexibility, often requiring manual adjustments by professionals and may not account for specific hardware or user preferences, leading to suboptimal initial fittings and potential need for frequent fine-tuning.
Innovation Solution
A deep neural network ensemble is trained on a dataset of existing hearing device fittings to predict a range of gain values for different frequency bands, allowing for personalized and flexible fitting settings, which can be used for both initial fittings and fine-tuning adjustments, enabling automated or semi-automated fitting processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional fitting rules are used for hearing devices, then the fitting process can be performed with existing methods, but the fitting lacks individualization and requires more manual fine-tuning time
Solution Approach 1:
The system performs preliminary actions by training a deep neural network on a comprehensive dataset of existing fittings before actual use. The trained model pre-calculates optimal gain values for new fittings based on user characteristics, eliminating the need for time-consuming manual fine-tuning later. This preliminary training phase enables the system to provide individualized fittings immediately without requiring subsequent adjustments.
2Extent of automation
If automated fitting processes are implemented for OTC hearing devices, then professional intervention is reduced, but the initial fitting quality may be compromised without manual expertise
Solution Approach 1:
The system copies the expertise and patterns from professional fittings by training the deep neural network on a large dataset of existing fittings performed by hearing care professionals. The model learns to replicate the decision-making process and gain value selections that professionals make, enabling automated systems to achieve comparable initial fitting quality without requiring manual expertise during the actual fitting process.
3Ease of manufacture
If proprietary fitting rules are used, then the fitting can be performed with manufacturer-specific guidelines, but the rules may not be up to date with new hearing device models
Solution Approach 1:
The system changes from static proprietary fitting rules to dynamic, data-driven parameters by training the deep neural network on a comprehensive dataset that includes fittings from multiple manufacturers and device models. The model learns to adapt to different hardware specifications and new device models automatically through the training data, eliminating the need for manual updates to fitting guidelines whenever new models are introduced.
4Measurement precision
If more data and complex models are used to improve fitting accuracy, then individualization improves, but the system complexity and computational requirements increase
Solution Approach 1:
The system replaces complex manual fitting procedures and iterative adjustment processes with a trained deep neural network model. Once trained, the model provides accurate fitting recommendations through automated computational processing, reducing the need for complex manual interventions and multiple adjustment sessions while maintaining high individualization and accuracy.
Data Source
AI summary
A method and a system for fitting a hearing device to a hearing loss of a user is devised. The method utilizes a data pool of existing hearing device fittings for training a deep neural network of the system to predict a dataset of gain value ranges for a hearing device fitting when the system is presented with a user profile comprising an audiogram, a set of user data, and a proposed hearing device. The dataset of gain value ranges predicted by the system may be subjected to statistical methods for providing a set of gain values suitable for being applied directly to the hearing device to be fitted. The system and the method may beneficially be used for automatically fitting an OTC hearing device, for checking a proposed hearing device fitting, or for updating an existing hearing device fitting, for instance when relevant data in the user profile changes.