Nonlinear And Multiple Parameters Research Of Sleep-staging

Sleep is a kind of important physiology phenomena. A good physiological sleephas the function of restoring physical energy and brain energy in human being. Asrapid rhythm of morden society and the increasing competion pressure, more andmore people have bad sleep quality, usually suffering from insomnia, which affectslife quality, work efficiency and health. Now sleep apnea syndrome could even bevital to life. Questions about Sleep quality are concerned and good judgementmethods are demanded urgently. In addition, some potential diseases, especially incerebra, are not easily founded when waking, but they will appear during sleep. So, ifwe can distinguish the difference between different sleep stages using some methods,it will be helpful to develop new medical measurement, preventability and treatment.In recent years, an increasing research effort has been directed to developing areliable method to determine the depth of sleep from electroencephalograph (EEG)signals. EEG is a kind of electricity physiological reaction generated by cerebralnervous system and can be mensured without hurt. The body of people is a systemwhich means that different sleep stage will also affect other physiological signals,such as ECG, BP, animal heat, respiration, EMG and so on. Researching the relationbetween those physiological signals and sleep or getting sleep information from thosesignals is a good complementarity to the sleep-staging based on EEG. Its find form thecalculation result of this paper that ECG and BP have better regularity than EEG andthe difference of BP and ECG in different stage is samller than EEG. So, taking otherphysiological signals into consideration when make sleep-staging will improve theaccuracy of staging which is the direction for future sleep-staging.In recent years, it's finds that the physiological signals of people are active,causal, time-varying, no-stationary and complicated noliear dynamics signals. Whichmake the algorithm based on time-domain and frequency domain incapable ofdisposing such signals. So three kinds of nonlinear methods are used here to process the sleep physiological signals at different stage. The entropy complexity of sampleentropy and multi-scale entropy and scale exponent of detrended fluctuation analysiswere used here to delineate the deep change in the wholly sleep course.The sample entropy has a good consistency compared with approximate entropywhich is also used widely in many field. It means the sample entropy have weakdependence on the length of the data and embedded dimension. When processing theEEG the sample entropy can't distinguish the S3 and S4 well. So wavelet is need hereto construct the signals in certain scale. And the result is well combining the entropyand multiple scale wavelet. The multiple scale entropy also take the multiple scale andthe the widely used sample entropy into consideration. And the sort effect is alsosatisfying. The detrendde fluctuation analysis used a different way which calculatesthe scale exponent of the sleeping physiological signals and different kind of signalsare have different scale exponent. The larger the scale exponent the stronger theregularity is for the signals. The outcome shows the scale exponent has abviousdifference in different stage of sleep, and the detrended fluctuation anaysis have abetter effect in sorting. We also find in the paper that using the same method used inprocessing the EEG sinals can't get a satisfying result in processing the ECG and BPsignals in the sleep stage. The chance is slim that just use the ECG and BP can alsoget a satisfying result but we can get that purpose by process as many physiologicalsignals as possible beside EEG..