Hearing Device Fitting Agent with Collaborative User Model Initialization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The traditional methods for fitting and tuning hearing devices rely heavily on audiograms and expert feedback, leading to dissatisfaction among users and frequent visits to healthcare professionals, as they do not consider individual user preferences effectively.
Innovation Solution
A fitting agent system that initializes a user model using a user preference function and response distribution, obtains test settings, presents them to the user for feedback, and updates the model based on preferred settings, incorporating environment data and reference user preferences to minimize interactions and optimize hearing device parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional audiogram-based fitting methods are used, then hearing device parameters can be configured, but user preferences are not considered leading to dissatisfaction and frequent visits
Solution Approach 1:
The system implements continuous feedback loops where user responses to test settings are captured and used to update the user model. The fitting agent presents multiple test settings, detects user preferences, and iteratively refines parameter recommendations based on this feedback, enabling remote tuning without requiring repeated in-person visits to healthcare professionals.
Solution Approach 2:
The fitting agent enables users to self-service their hearing device configuration by independently evaluating test settings and providing preference feedback. The system autonomously processes user responses, updates the user model, and generates optimized parameter recommendations without requiring continuous HCP intervention, thereby reducing visit frequency while maintaining satisfaction.
2Adaptability or versatility
If an interactive agent learns user preference from scratch, then personalized settings can be achieved, but many interactions are required
Solution Approach 1:
The system performs preliminary actions by pre-initializing the user model with collaborative priors derived from aggregated data of reference users with similar profiles. This pre-learning phase incorporates environmental probabilities and group preferences before the individual user begins interacting, providing a head start that reduces the number of interactions needed for accurate personalization.
Solution Approach 2:
The system introduces collaborative priors as an intermediary between generic fitting rules and individual user preferences. These priors, derived from reference user groups, serve as a bridge that provides personalized starting points without requiring extensive individual interaction data, thereby reducing the time needed to achieve accurate personalization.
3Measurement precision
If collaborative priors from reference users are used, then initialization is improved, but user privacy and data security concerns arise
Solution Approach 1:
The system extracts only the necessary aggregated statistical information (collaborative priors, environmental probabilities, preference distributions) from reference user data while leaving out individual identifiable information. By separating the useful pattern recognition from personal data, the system achieves accurate model initialization without compromising user privacy or requiring storage of sensitive personal information.
Data Source
AI summary
A fitting system includes one or more processors configured to: initialize a user model; obtain a test setting comprising a primary test setting and a secondary test setting; output the primary test setting and the secondary test setting for presentation to a user; obtain a user input for a preferred test setting indicative of a preference for either the primary test setting or the secondary test setting; and update the user model based on a hearing device parameter of the preferred test setting; wherein the one or more processors are configured to initialize the user model by: obtaining a profile of the user; obtaining a group of reference users; obtaining reference posteriors of reference users in the group of reference users; determining a collaborative user preference distribution based on the reference posteriors; setting the collaborative user preference distribution as a prior; and initializing the user model based on the prior.