Smartphone-Based Objective Hearing Disorder Screening

BACKGROUND:

Hearing disorders are widespread, but many current screening and diagnostic approaches depend on subjective patient responses or expensive clinic-based equipment. This can limit accessibility for children, older adults, cognitively impaired patients, rural populations, and individuals who cannot reliably complete conventional hearing tests.

TECHNOLOGY OVERVIEW:

University at Buffalo technology AudioSight combines a smartphone application, a custom 3D-printed smartphone attachment, near-infrared illumination, and cloud-based artificial intelligence analysis. The attachment includes a 15× macro microlens, concentric 730 nm near-infrared LEDs, and a soft eye cup designed to reduce ambient light and stabilize pupil imaging. The prototype hardware is estimated to cost less than $20 per unit in small-batch production.

The AI analysis pipeline uses a TemporalUNet model for pupil segmentation, XGBoost-based blink detection and artifact correction, and feature extraction from the pupil-response curve, including response latency, peak and valley amplitudes, dilation velocity, area under the curve, and recovery time. A BiLSTM-attention model called RiskNet combines these features with clinical information to generate risk scores for auditory dysfunctions including hidden hearing loss, tinnitus, and hyperacusis.

ADVANTAGES:

• Non-volitional measure of auditory function using the audio-evoked pupil response rather than relying only on self-report
• Low-cost smartphone-based hardware
• Designed for self-administration, and potentially suitable for children, older adults, cognitively impaired patients, and underserved populations

APPLICATIONS:

• At-home screening for hearing-related disorders
• Screening for hidden hearing loss, tinnitus, and hyperacusis
• Teleaudiology and remote patient monitoring

INTELLECTUAL PROPERTY SUMMARY:

Copyright for code and UI.

STAGE OF DEVELOPMENT

TRL 6

LICENSING STATUS:

Available for licensing or collaboration.