Figure 8. .

Relationship between the interindividual variations of the empirical graph-theoretical metrics and modeling results for different parcellations. (A) Amounts of within-parcellation, between-subject variance in the modeling results (goodness-of-fit to empirical data) being explained via multivariate ordinary least squares linear regression utilizing the z-scored graph-theoretical statistics of the empirical connectomes per parcellation. Modeling results were sampled by using the systems of coupled phase oscillators (red) and neural mass models (blue). (B–C) Regression coefficients corresponding to the data variables (network properties) depicted in Figure 2 and Figure 3 for four selected brain parcellations as indicated in the legend and for the phase oscillators (B) and the neural mass models (C) leading to the regression results in panel A. The abbreviations MIST (103), Shen (156), Sch. (100), and EK (86) correspond to the parcellations in Table 1 and in panel A with indices 3, 10, 11, and 16, respectively. (D) Pearson correlation coefficients across the regression coefficients per pair of brain parcellation and model type. Table 1 contains the parcellation information corresponding to the atlas indices. Abbreviations: coef. = coefficient, corr. = correlation.