Fig. 1

A: Autocorrelation function from MEG timeseries. Each autocorrelation function is a mean of autocorrelation functions of all the brain regions, and therefore each line corresponds to a single subject. The timescale is the value of lag that corresponds to an autocorrelation function value of \(e^{-1}\). B: Distribution of timescales obtained from autocorrelation function. Based on a Kolmogorov-Smirnov test, time constant of MEG recordings for patients with AD is significantly larger than the timescale of MEG recordings for healthy controls (\(p<0.001\), Cohen’s D effect size = 0.42). C: Power spectral density (PSD, in dB scale) extracted from the MEG recordings. Each PSD is a mean of all regions and then centered to the mean and scaled to unit variance for every subject separately. D: Center frequencies of the first and the second peaks of the PSD in C for every subject. Center frequencies of both the first and the second peaks are lower in AD (\(p<0.001\), Cohen’s D effect size = 0.21 for the first peak, \(p<0.001\), Cohen’s D effect size = 0.32 for the second peak). The peaks were extracted using the FOOOF toolbox [31]