Ear-worn reference data collection and annotation for multimodal context-aware hearing instruments.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Human-Computer Interaction

Background:

• Developing advanced hearing instruments (HIs) requires unobtrusive methods to capture real-world user behavior.
• Existing methods for capturing head movements may be intrusive or limited in real-life applicability.

Purpose of the Study:

• To introduce a novel ear-worn sensing and annotation device for unobtrusively capturing head movements.
• To demonstrate the utility of head movement data for enhancing multimodal hearing instruments.

Main Methods:

• Development of an ear-worn device capturing triaxial acceleration, rate of turn, and magnetic field.
• Implementation of a one-button interface for real-time user annotation.
• User study with 21 participants to assess comfort, device hold, and system performance (5+ hours runtime at 128 Hz).
• Application of unsupervised clustering on head acceleration data.

Main Results:

• The device was perceived as comfortable and demonstrated robust hold during the user study.
• Unsupervised clustering successfully utilized head movement data, showcasing its benefits for multimodal HIs.
• The system achieved a runtime exceeding 5 hours at a 128 Hz sampling rate.

Conclusions:

• The developed ear-worn device offers a practical solution for unobtrusively capturing head movements in real-world settings.
• Head movement data is beneficial for advancing the capabilities of multimodal hearing instruments.
• This novel technology has the potential to significantly improve hearing instrument technology.