In-Home Diagnosis of Obstructive Sleep Apnea Using Automatic Video Analysis

Abstract

Study objectivesTo evaluate the diagnostic accuracy of a non-invasive technology based on image processing for the identification of obstructive sleep apnea (OSA) and its severity at patients’ home. MethodsObservational, prospective, diagnostic accuracy study to evaluate the degree of measure agreement between Sleepwise (SW), in-laboratory attended polysomnography (PSG) and a home sleep apnea test (HSAT). 38 consecutive subjects with suspected OSA referred as outpatients to the sleep unit were recruited from September 2016 to September 2017. All patients underwent in-laboratory attended PSG and image processing with SW simultaneously overnight. Subsequently, a HSAT and image processing with SW were performed simultaneously overnight at patients’ home, and the 2 nights after, patients underwent only image processing with SW consecutively. ResultsIn-laboratory polysomnography and SW had a Lin's concordance correlation coefficient of 0.933 and a κ of 0.930. Between HSAT and SW the Lin's concordance correlation coefficient was 0.842 and a κ of 0.571. Agreement between two consecutive nights with SW recording showed a Lin's concordance correlation coefficient of 0.923 and a κ of 0. 837. ConclusionsSW was highly accurate for non-invasive and automatic diagnosis of OSA in outpatients compared to standard methods for OSA diagnosis either in-laboratory attended PSG or HSAT. SW proved to be a technique with repeatable and concordant results on different nights for the same patient. We conclude SW is a non-invasive, easy-to-use, portable, effective and highly accurate system for the in-home diagnosis of OSA.