The current study has some limitations. Firstly, we acknowledge the relatively small sample size, but these changes suggest an overall decoupling between the frequency bands even in the early stages of AD. Furthermore, we found a significant difference in the age between HC and MCI and have therefore used age as a covariate in the analysis. In addition, the follow-up time was short and, according to previous studies, the annual clinical progression rate is 15% (Petersen et al., 1999; Saxton et al., 2009), which means that only 30% of the patients with MCI should have progressed to AD in the present cohort. However, we found that 48% progressed, which may in part be due to the patients with MCI being at a more advanced stage of the disease at inclusion. Furthermore, we concatenated 1-s epochs and, due to the discontinuity, this may have induced some additional noise to the calculations. Due to line noise, we calculated the amplitude of gamma between 30 and 40 Hz, which is a narrow range compared to the previous studies analyzing CFC (Canolty and Knight,
