Features extraction from respiration rate variability signals for apnea prediction

Sleep can be expressed as an active process that circadian rhythm is entegrated with nervous system. Sleep disorders may appear during the sleep. Apnea, which is defined as the respiration stops more than 10 seconds during sleep, is the most important problem among the sleep disorders. In this study, prediction of apnea has been investigated statistically using features extracted from respiratory rate variability signals derived from respiratory signals recorded by polysomnography device during sleep. In order to detection of inspiration peaks in respiration signals taken by nasal cannula, Teager Energy Operators (TEO), threshold and multiple threshold algorithms have been used. The multiple threshold algorithm which gives the best results for the calculation of duration between peaks in respiration rate variability (RRV) signal. By using a GUI (Graphical User Interface) designed by MATLAB platform, 3 patients' all nasal cannula signals each of which contains 30 seconds duration epochs have been examined. Maximum, minimum and mean respiration rates, means, variances and standard deviations have been calculated for each epohcs of every patients. According to results, mean respiration rate and mean RRV calculated over the five epochs before the apnea have been found statistically important. As conclusion, these two parameters can be used for the prediction of apnea.