ABSTRACT Obstructive Sleep Apnea (OSA) is a prevalent disorder that remains undertreated due to poor adherence with continuous positive airway pressure (CPAP). Several studies suggest that the pharyngeal airway can collapse at different sites in patients with OSA (velum, oropharyngeal lateral walls, tongue base, and/or epiglottis). Furthermore, many of the non-CPAP treatments for OSA have been shown to work better for some sites of collapse than others. The problem is, the site of collapse requires expensive or invasive procedures, such as drug-induced sleep endoscopy (DISE). This proposal seeks to solve that problem by determining the site of collapse from acoustic analysis of snoring sounds. In Aim 1, a model for predicting the site of collapse will be developed using DISE with simultaneously measured snoring sounds in 800 individuals. In Aim 2, natural sleep endoscopy will be performed in a subset of people from Aim 1 to confirm that the collapsing structure(s) produce the same sound in natural sleep as drug-induced sleep. Finally, Aim 3 will test the ability to measure snoring sounds in the home environment. Comparisons will be made between snores from natural sleep endoscopy and natural sleep at home (in the same individual). Additionally, snoring will be measured at home on separate nights to test reproducibility across nights. The studies in this grant are expected to produce an algorithm for estimating the site of pharyngeal collapse from snoring sounds. This algorithm could improve the selection of patients for non-CPAP therapies and potentially increase the number of patients effectively treated for OSA.