Network-based methodology to determine obstructive sleep apnea

Abstract

Obstructive sleep apnea (OSA) is a long-term condition that often leads to severe problems on a personal and social level. Polysomnography is widely used as the most reliable method for diagnosing OSA, whereby the thermistor flow signal captures the respiratory dynamics and is used to assess respiratory differences between healthy and unhealthy subjects. To aid visual inspections of these signals and subsequent diagnostics, we here introduce a new method to study OSA using complex network theory. In particular, we first construct networks from thermistor flow signals and then determine their clustering coefficient and permutation entropy. We show that both quantities are useful discriminators between healthy and ill subjects. We provide accurate statistics between the control group and the OSA group, based on which we conclude that the proposed methodology is suitable for reliably determining OSA.