A Bone-Conduction Transducer-Based Detection System for Sleep Apnea Screening in the Family Units

In order to reduce the burden of medical PSG testing, we proposed a system or a way for preliminary screening of OSA and rehabilitation self-examinations for patients with moderate or severe OSA in family units. The snoring vibrations of 28 sets of data were collected by this system. And then, performing segmentation processing of the acquired signals in the time domain to obtain the snore silence segments, and two preliminary screening methods of OSA were proposed based on the silence segments, which were the SSI-Diagnostic Method and
$\xi $
-Diagnostic Method. All data were divided into 4 groups, based on PSG Apnea Hypopnea Index (AHI): No OSA, (n = 5;AHI< 5), mild (n=6; 5 ≤ AHI< 15), moderate (n=4; 15 ≤ AHI< 30) or severe OSA (n=13; AHI>30 events/h). The thresholds of the
$\xi $
-diagnostic method were obtained from the
$\xi $
-ROC scatter plots, 0.0236, 0.0666, and 0.2134 for simple/mild, mild/moderate, and moderate/severe OSA, respectively. Bland–Altman graphical analysis showed the strong agreement between SSIs and AHIs (r = 0.96, p < 0.001). Additionally, results showed that, except for NPV, all indicators are greater than 90%, the
$\xi $
-diagnostic method had a more balanced, real and excellent preliminary screening performance than the SSI-diagnostic method among the diagnostic indicators involved. The preliminary screening performance of proposed screening system with particularly significant practical significance could meet the screening requirements of potential OSA patients and self-examinations of rehabilitation training for much heavier OSA patients in family units to reduce the burden of screening and PSG testing of patients.