Sleep-Wake Detection based on Respiratory Signal Acquired through a Pressure Bed Sensor

Abstract

This study proposes an automatic method for the sleep-wake staging in normal and pathologic sleep based only on respiratory effort acquired from a Pressure Bed Sensor (PBS). Motion and respiratory movements were obtained through a PBS and sleep-wake staging was evaluated from those time series. 20 all night polysomnographies, with annotations, used as gold standard and the time series coming from the PBS were used to develop and to evaluate the automatic wake-sleep staging. The database was built up by: 10 healthy subjects and 10 patients with severe sleep apnea. The agreement of the statistical measures between the automatic classification and the human scoring were: 83.59 +/- 6.79 of sensitivity, 83.60 +/- 15.13 of specificity and 81.91 +/- 6.36 of accuracy. These results suggest that some important indexes, such as sleep efficiency, could be computed through a contact-less technique