Connectome‐based predictive modelling of cognitive reserve using task‐based functional connectivity

Figures

• 

  FIGURE 1. Illustration of CR residuals from the regression of global cognition on age, sex, and brain structure. Positive residuals (green arrows) reflect better cognitive performance than expected given age, sex, and brain structure. Negative residuals (purple arrows) reflect poorer cognitive performance than expected. Higher/more positive residual values reflect higher CR. Image adapted with permission from Figure 1 (Franzmeier, Hartmann, et al., 2017)

• 

  FIGURE 2. Schematic of CPM with LOOCV to predict CR residuals in the training set. PosNetStrength = positive network strength; NegNetStrength = negative network strength; ComboNetStrength = combined network strength; PosNet CR = positive network strength predicted CR; NegNet CR = negative network strength predicted CR; ComboNet CR = combined network strength predicted CR

• 

  FIGURE 3. CR residual versus positive‐, negative‐, and combined network strength predicted CR in the training set

• 

  FIGURE 4. NART scores versus positive‐, negative‐, and combined network strength predicted CR in the training set

• 

  FIGURE 5. CR residual versus positive‐, negative‐, and combined network strength predicted CR using adjusted CPM in the training set

• 

  FIGURE 6. NART scores versus positive‐, negative‐, and combined network strength predicted CR using adjusted CPM in the training set

• 

  FIGURE 7. Circle plots illustrating the positive and negative CR connectomes. Positive connections (red) and negative connections (blue) in original CPM (top panel) and adjusted CPM (bottom panel), controlling for age, sex, and mean FWD. These circle plots are inverted such that the right side of each plot corresponds to the left hemisphere and the left side to the right hemisphere.

• 

  FIGURE 8. Glass brain visualising the patterns of connectivity within the brain. Positive connections (red) and negative connections (blue) in original CPM (a, top panel) and adjusted CPM (b, bottom panel), controlling for age, sex, and mean FWD

• 

  FIGURE 9. Connectivity matrices summarising the connectivity patterns within and between different functional networks. Note: Darker shades represent stronger connectivity (i.e., larger number of edges in that network). MF = medial frontal network; FP = Frontoparietal network; DMN = default mode network; mot = motor network; Vis I = visual I network; Vis II = visual II network; VAs = visual association network; SAL = salience network; SC = subcortical network; CBL = cerebellar network

• 

  FIGURE 10. CR residual versus positive‐, negative‐, and combined network strength predicted CR in the test set. a = predicted CR values from original network strength models; b = predicted CR values from adjusted network strength models, controlling for age, sex, and mean FW

• 

  FIGURE 11. NART scores versus positive‐, negative‐, and combined network strength predicted CR in the test set. a = predicted CR values from original network strength models; b = predicted CR values from adjusted network strength models, controlling for age, sex, and mean FWD