Objective response detection in the frequency domain: comparison of several q-sample tests

Area of Science:

• Auditory Neuroscience
• Statistical Signal Processing
• Hearing Science

Background:

• Objective assessment of hearing thresholds is crucial for diagnosing hearing loss.
• Auditory-evoked potentials (AEPs) provide objective measures of auditory system function.
• Detecting faint AEP signals requires robust statistical methods for response identification.

Purpose of the Study:

• To evaluate the performance of different q-sample statistical tests for detecting responses in auditory-evoked potential spectra.
• To identify the most powerful statistical test for objective hearing threshold assessment.

Main Methods:

• Extensive Monte Carlo simulations were used to assess the performance of three q-sample tests: Watson's U2 analogue, uniform scores test, and a modified uniform scores test.
• Sensitivity and receiver-operating characteristics (ROCs) were calculated to compare test performance.
• Real near-threshold auditory brainstem response data was used for validation.

Main Results:

• The modified q-sample uniform scores test demonstrated superior performance compared to the other tested methods.
• Monte Carlo simulations and real data analysis confirmed the enhanced power of the modified test.
• Receiver-operating characteristics analysis provided a clear comparison of sensitivity and specificity.

Conclusions:

• The modified q-sample uniform scores test is recommended as the most powerful statistical tool for objective hearing threshold assessment.
• This test offers improved response detection in auditory-evoked potential analysis.
• Practical applications of this superior statistical method in audiology are discussed.