Results
Figure 1 illustrates the EEG power and FC analysis for a representative subject. Figure 1a shows the topoplot of raw power in the slow gamma range between 250-750 ms after stimulus onset, while Figure 1b shows the change in power from baseline (-500 to 0 ms), revealing an increase in power in the occipital area. As in our previous studies, all analyses were performed on three groups of occipital electrodes (left, right and back; see Methods for details). Figure 1c shows the change in average power spectral density profile between stimulus and baseline periods for the left (top) and right (bottom) electrode groups (these electrodes are highlighted in Figure 1b), revealing a clear suppression in alpha and elevation in slow and fast gamma bands. Consistent with our previous studies, we limited analyses to the above listed frequency bands. To estimate FC, we computed PPC between each of the electrodes in the left, right or back group as “seed” and all other electrodes. Subsequently, we averaged the PPC maps for all electrodes within a group to generate an average PPC map for each group. Figure 1d shows the average PPC maps with seed electrodes on the left (top) and right (bottom), which we aim to contrast across different age groups and disease conditions.
The maps in Figure 1d show that FC falls with increasing inter-electrode distance, as expected. However, some electrodes may have higher FC with respect to the seed electrodes than what is expected based on their inter-electrode distance. For example, left and right group of seed electrodes may show enhanced FC due to potential inter-hemispheric synchrony (Engel et al. , 1991). Figure 1e shows the FC for different electrodes (frontal, central, temporal and occipital; see inset for details) as a function of the inter-electrode distance (see Methods for details) from each one of the seed electrodes in the left (top) or right (bottom) electrode group. The FC decrease with inter-electrode distance did not show any obvious deviations based on the brain region. In particular, the homologous inter-hemispheric occipital electrodes (yellow triangles and circles for top and bottom plots, respectively) did not show greater FC as compared to intra-hemispheric electrodes in the central region (red circles and triangles for top and bottom plots, respectively). Further, different seed electrodes within a group (different columns in Figure 1e) showed similar trends. We therefore considered FC values between electrodes pairs solely based on inter-electrode distance, ignoring which electrode group they belonged to (i.e., ignoring the color and symbol in Figure 1e), and further pooled across different seed electrodes (pooling data across the three columns in Figure 1e). Two more example subjects are shown in Supplementary Figure 2 and 3.
To compare how FC varies with age, we divided the population into two groups – middle-aged (50-64 years; N=91) and elderly (>64 years, N=127), as done in our previous study (Murty et al. , 2020) and other studies (Sanderson & Scherbov, 2008). Splitting continuous variables, like age in our present study, into artificial groups is usually done for simplicity in interpretation and visualization but may lead to problems such as reduced statistical power, inefficacy of corrective measures for accounting confounding factors, incompatibility across the studies, non-existent associations and unreliable risk estimates (Naggara et al. , 2011). We therefore first show results using grouped analysis to be comparable with previous studies (Figure 2-5), and subsequently confirm our results using regression analysis (Figure 6). Indeed, as shown later, regression analysis reveals some important associations that are not observed in grouped analysis.