An analytical technique to evaluate early auditory evoked potential morphology - PROJECT SUMMARY / ABSTRACT Diagnostic tests for hearing loss miss many underlying sensory and neural pathologies that contribute to hearing problems. Treatments such as hearing aids and cochlear implants are prescribed without knowledge of the underlying site-of-lesion or remaining function, resulting in large variability in benefit and satisfaction. Furthermore, undetected and unaccounted for pathologies may be a cause of the mixed success in current clinical trials of pharmaceutical treatments for hearing loss. Early-latency auditory evoked potentials (AEPs) are an objective measure of auditory physiology that could identify and differentiate sensory and neural pathologies. Despite the clinical availability of early-AEPs, there are barriers to their widespread use that we argue largely result from reliance on visual inspection of features in the evoked response waveform to identify pathology. Reliable analysis using visual inspection requires significant training and more time than is customary for a typical hearing evaluation appointment. Reliable interpretation requires experience that is too burdensome for a non- specialist. A recently developed technique overcomes these barriers by automatically analyzing the response waveform and extracting more features than can be seen by the human eye. Though the preliminary version of the technique's analysis is automatic, it requires some manual adjustment beforehand to work on each new dataset. Furthermore, there is minimal existing research on the benefit of the additional extracted features analyzed by the technique in their ability to identify certain auditory pathologies. Therefore, the proposed study has two aims. The first aim automizes the adjustment process of the technique such that it works immediately on early-AEPs from different species, stimuli, and recording parameters. The second aim tests whether the features analyzed by the technique can better identify several sensory and neural pathologies in animals than the traditional method of visual inspection. To achieve these aims, existing AEP responses from human and animal research labs will be collected. The technique will be trained to work automatically on healthy human and animal waveforms, similar to training artificial intelligence, with an expected outcome being a user-friendly technique that requires minimal input (e.g., species and stimulus rate) to automatically analyze responses. For the second aim, several animal models of different sensorineural diseases will be used to compare the technique to visual inspection and other automated approaches in their ability to identify diseased ears from healthy controls. The proposed study is the next step on the path to creating a no-cost user-friendly tool for automated analysis and assistance in interpreting features to identify disease. This effort is needed to work toward a subsequent R01 translating animal morphological biomarkers to human early-AEPs. The long-term goal of this line of research is to identify pathologies underlying sensorineural hearing loss so that new treatments can be developed that target specific pathologic mechanisms in the ear.
