Project 2 Summary - Abstract Since the inner ear cannot be biopsied, and clinical imaging cannot produce cellular-level resolution, the only way to uncover the functionally important structural changes underlying sensorineural hearing loss is the microscopic examination of post-mortem human temporal bones. This P50 Center focuses on primary neural degeneration in the inner ear (i.e. the loss of synaptic connections between surviving sensory cells and auditory nerve cells), and its hypothesized role in limiting the ability to understand complex stimuli like speech and as a key elicitor of tinnitus and hyperacusis. Over the past 5 years, Project 2 showed, as predicted from animal models, that the rate of auditory nerve loss in normal-aging human ears out-paced the rate of hair cell loss by 2:1, and that this neural loss was further accelerated in those with a history of noise exposure. A rigorous statistical model showed that while the audiometric thresholds were well predicted by the patterns of hair cell loss, the neural loss did not affect threshold but contributed to the differences in word identification abilities among those with similar audiograms. Over the next 5 years, Project 2 builds on this foundation to explore a wider range of acquired hearing loss etiologies, i.e. ototoxic antibiotics and chemotherapeutics, sudden sensorineural hearing loss and Ménière’s disease. We choose this because they are common, well represented in our temporal bone archives, and are often associated with tinnitus and difficulties in speech discrimination. We are developing machine-learning approaches to automate the acquisition of quantitative histopathological data, because we aim to grow the number of cases analyzed as rapidly as possible. As we have now entered the era of clinical trials for deafness therapeutics, there is a critical need for robust statistical models to accurately predict the degree and pattern of cellular loss from the audiogram, word score, hearing loss etiology, age and sex of a candidate patient.