Profiling hearing aid users through big data explainable artificial intelligence techniques

Area of Science:

• Audiology and Hearing Science
• Artificial Intelligence in Healthcare
• Biomedical Data Science

Background:

• Hearing loss (HL) significantly impacts quality of life, with low adoption and satisfaction rates for hearing aids (HAids).
• Understanding factors influencing HAid success is crucial for improving patient outcomes and reducing societal burden.
• Modern HAids offer dynamic logging capabilities, presenting an opportunity for data-driven insights.

Purpose of the Study:

• To describe the data collection process and framework for comprehensive patient profiling in the SMART BEAR project.
• To identify factors influencing hearing aid satisfaction, usage, and overall rehabilitation success.
• To leverage AI and machine learning for predicting patient behavior and optimizing interventions.

Main Methods:

• Collecting 12-month continuous monitoring data from hearing loss patients, including demographics, audiometric, medical, cognitive, and behavioral data.
• Utilizing data from medical devices, wearables, and clinical assessments (face-to-face and remote).
• Applying descriptive and AI-based analysis, including time-dependent machine learning and Explainable AI (XAI).

Main Results:

• The study proposes a framework for creating detailed patient profiles for individuals with hearing loss.
• AI-driven analysis will identify correlations between heterogeneous data and hearing loss parameters.
• Predictive models will forecast patient satisfaction and HAid usage patterns.

Conclusions:

• Comprehensive patient profiles derived from integrated data sources can significantly enhance hearing rehabilitation.
• Identifying factors associated with HAid dropout, usage duration, and satisfaction is key to personalized treatment.
• The SMART BEAR framework aims to optimize interventions, reduce healthcare costs, and improve patient quality of life.