A gender-aware framework for the daytime detection of obstructive sleep apnea.

Abstract

Sleep is an activity that is necessary for our survival. While the body may be still during sleep, the brain is actively progressing through repeating cycles of light and deep sleep whose purpose is physical and mental recovery and regeneration. Obstructive sleep apnea (OSA) is a sleep disorder in which breathing is frequently and repeatedly stopped during sleep. OSA severely interrupts the normal sleep cycle and the regeneration work associated with it and can thus result in detrimental health consequences. OSA, with all the adverse health effects associated with it, places a significant burden on the US healthcare system. Polysomnography (PSG) - the gold standard OSA diagnostic test - is an overnight sleep test that monitors the biophysiological changes that occur during sleep. The test is notorious for its intrusiveness, discomfort, prohibitive cost, and scarcity - all reasons contributing to OSA being a severely underdiagnosed sleep disorder. In this paper, we propose a system that can serve as an early-stage OSA diagnostic tool that can non-intrusively, affordably and accurately screen patients for the disorder before proceeding with a full-night PSG. Unlike existing tools, our solution is gender-aware and does not rely on detecting apneic events in the data to make a diagnosis; rather, it is designed to trigger brain responses that are indicative of the disorder. Our tool can therefore make diagnoses even while patients are awake and breathing normally. The system was tested in a pilot study of 21 patients and our preliminary results show an average accuracy of 96.25%.