Alzheimer's disease (AD) is a neurological degenerative disease, which is mainly char-acterized by the memory loss. As electroencephalogram (EEG) device is relatively cheap, portable and non-invasive, it has been widely used in AD-related studies. We proposed a method to detect the differences between healthy subjects and AD patients, which combines classical sample entropy (Sam-pEn) and surrogate data method. EEGs from 14 AD patients and 20 healthy subjects were analyzed. The results based on the original data showed that the SampEn of AD patients was significantly de-creased (p < 0.01) at electrodes c3, f3, o2 and p4, which confirmed that AD could cause complexity loss. However, using original data could be subject to human judgement, so we generated a series of surrogate data. We found that, there were significant difference of SampEn between the original time series and their surrogate data at c3 and o2 electrodes and the differences between healthy subjects and AD patients can be verified. Our method is capable of distinguishing AD patients from healthy subjects, which is consistent with the concept of physiologic complexity, and providing insights for understanding of AD.
Keywords: Alzheimer’s disease; electroencephalogram; nonlinear time series; sample entropy; surrogate data analysis.