[Clinical views on biosignal processing of ventilation in sleep medicine]

Biological signals recorded by polysomnography such as electroencephalograms (EEG), electrooculograms (EOG) und electromyograms (EMG), airflow, thoracic and abdominal motions, O(2) saturation and heart frequency pass through a processing chain with amplification and filtering of the measured variables. In the signal processing the original signal is manipulated to extract information in order to correlate various parameters and to visualise the signal in different representations. The task of the clinician is now to take advantage of the information available and to come to clinical conclusions. Therefore the characteristics, especially the methodology of the signal processing in the devices, should be disclosed by the manufacturers and thus be verifiable. In this review for the example of O(2) saturation, the consequence of different processing criteria with regard to the apnea/hypopnea index is presented. Biosignals such as the respiratory flow are also increasingly being used to control therapy equipment for automatic CPAP devices (APAP). However, the implemented algorithms are often not known, and cannot directly be derived from the polysomnographic records under APAP therapy. In model experiments with a flow generator, defined flow patterns can be simulated. Using bench tests with simulated apneas and waking periods, diagnosed from the flow characteristics, the methods of signal processing can be verified by measuring the pressure profiles and made more transparent for the clinician.