Calibration offset estimation in mobile hearing tests via categorical loudness scaling

Area of Science:

• Audiology
• Mobile Health Technology
• Signal Processing

Background:

• Smartphone-based hearing assessments offer accessibility but require accurate device calibration.
• Calibration offsets can significantly impact the reliability of hearing test results.
• Categorical loudness scaling (CLS) provides level-independent measures robust to calibration errors.

Purpose of the Study:

• To develop and evaluate methods for estimating device calibration offsets in smartphone-based hearing assessments using CLS.
• To enable reliable and accurate hearing evaluations using mobile devices in diverse environments.
• To improve the accessibility of hearing healthcare through technology.

Main Methods:

• Simulated calibration offsets using a Gaussian distribution.
• Trained Bayesian regression and nearest neighbor models on CLS parameters and Oldenburg Hearing Health Record (OHHR) data.
• Utilized CLS for its robustness to calibration inaccuracies.

Main Results:

• The Bayesian regression model achieved a median absolute error of approximately 5 dB for estimating calibration offsets.
• Calibration uncertainty was reduced by factors ranging from 0.41 to 0.79.
• Demonstrated increased robustness of hearing assessments in uncontrolled environments.

Conclusions:

• CLS-based models show potential for compensating missing calibration in simulated smartphone hearing tests.
• Further validation with real-world listener data is necessary.
• This method offers a practical alternative to traditional threshold-based approaches, enhancing hearing healthcare accessibility.