A novel detection method for obstructive sleep apnea based on wavelet information entropy spectrum

Accurately detecting and judging apnea make great significance to the diagnosis of obstructive sleep apnea (OSA) using different types of detection technologies, such as contact or non-contact technology, which is meaningful for the evaluation and treatment of OSA patient. However, the difference of deriving means from different detection technologies for the data and selected features make great influence on the performance of a variety of commonly used OSA detection algorithms. This paper proposed a novel apnea detection method based on wavelet information entropy spectrum. The internal feature derived from the strong irregularity, complex composition and disorder of apnea signal was exploited to distinguish the apnea case. According to the apnea recognition experimental results for the bio-radar and PSG respiratory signal using this novel method, the apnea judgment accuracy of the novel method for bio-radar signal is 93.1%, while that for PSG signal could even reach 96.1%. It demonstrates that this method manifests excellent robustness for both the non-contact and the contact detected respiratory signal while guaranteeing high judgment accuracy.