Research on Sleep EEG Time-Series Using Nonlinear Sample Entropy
Ming‐Shi Wang · Dianzi qijian · 2008
Compared sample entropy algorithm with approximate entropy algorithm. The two kinds of entropy analysis of a mixture system including stochastic signal and deterministic signal show that sample entropy is better than approximate entropy in time-series complexity analysis when the tolerance threshold lower than 0.2 requirement. And then, the experiment data of different sleep stages acquired from one night sleep EEG signal was analyzed using sample entropy as a characteristic value. It is clearly that the sample entropy of different sleep stages is difference, the deeper the sleep the smaller the sample entropy. Therefore, sample entropy can divide sleep stages well and can be an important nonlinearity characteristic parameter in sleep automatic stage.