Abstract
In this contribution, we investigate the effectiveness of using filtering techniques based
on blind source separation methods for continuous monitoring of input respiratory impedance to
detect obstructive sleep apnea (OSA). The simple and non-invasive forced oscillation lung function
test is used to apply respectively a 1Hz and a 3Hz input pressure oscillation to the patient during
spontaneous ventilation. Once the breathing signal extracted by means of independent component
analysis (FastICA) method, the input respiratory impedance can be obtained by the ratio between
respiratory pressure and flow at the excited frequency. This index is then used to detect changes in
the respiratory impedance values as a result of the apneic events. The advantages of the blind
source separation techniques are emphasized and its use to assess respiratory mechanics for
detecting OSA episodes is extracted from this study.
