Audiogram Estimation Performance Using Auditory Evoked Potentials and Gaussian Processes.

The study was conducted in awake adults, whereas the primary clinical target for this technology is sleeping infants or uncooperative patients. The authors explicitly note that test time reductions may not generalize to a clinical setting. The visual inspection benchmark was performed by audiology undergraduates with limited experience, not highly trained professionals. This likely overestimates the examiners’ test time and underestimates their accuracy compared to real-world clinical performance. The British Society of Audiology guidelines used by examiners were designed for sleeping infants and were too strict for noisy adult data, requiring modifications that may have further disadvantaged the visual inspection group. The GP algorithm’s active learning rules and stopping criteria were not fully optimized and may be suboptimal. The study did not evaluate the method’s ability to detect abnormal test conditions or artifacts, which is a known risk in fully automated systems.

The work was funded by the William Demant Foundation. All authors contributed to the study’s conception and algorithm design. The funder is associated with hearing aid technology, though the authors state they contributed to the conception and design.

Do not assume this method is ready for clinical use with infants or uncooperative patients. The study validated the algorithm in awake adults, and the authors state that the test time reductions may not generalize to a clinical setting. Furthermore, the comparison against visual inspection used inexperienced undergraduates following modified guidelines, which likely exaggerated the method’s apparent efficiency advantage over trained clinicians.