On-line detection of sleep apnea during critical care monitoring

If sleep apnea is recognized in time, the number of apneal breaths can be decreased or even fully eliminated. If the patient is artificially respirated, then a solution could be the appropriate control of the respirator: the compressor creates air pressure that is directed through the airway, keeping it open to allow the patient to sleep and breathe normally. Another option can be the use of various medications. The paper presents an on-line automated method for the detection of sleep apnea. We found important the use of data fusion technology, which fuses all available respiration signals (in our case nasal and abdominal), and an additional verification signal obtained from another source (blood oxygen saturation). The detection proceeds in two stages. The first stage of signal processing contains separated detectors for each input signals. The outputs of these detectors are the features which are required for apnea detection. The second stage of signal processing realizes a data fusion unit based on empirical rules. The output of the system is the apnea alarm signal.