Fig. 2: Normative models for WM microstructure (fractional anisotropy: FA) and gray matter (GM) volume and their relationships with d factor score estimates. | Nature Communications

Fig. 2: Normative models for WM microstructure (fractional anisotropy: FA) and gray matter (GM) volume and their relationships with d factor score estimates.

From: Neural–genetic–environmental evidence for a disease factor in mental and physical health multimorbidity

Fig. 2: Normative models for WM microstructure (fractional anisotropy: FA) and gray matter (GM) volume and their relationships with d factor score estimates.

a For each sex, normative models of FA in WM tracts and GM volume in brain regions were separately constructed. Only the t values of regions that survived Bonferroni correction (p < 0.05) in the multiple linear regression analyses between regional z-scores and the d factor scores are shown. Age-related trajectories are plotted for the WM tract and brain region with the largest absolute t values in males and females, respectively (red = female, blue = male). Individual patient data points are plotted as semi-transparent scatter points. The central line represents the median predicted value from the healthy normative model, and shaded bands indicate the 25th–75th and 5th–95th percentile ranges. Wider bands correspond to increased epistemic uncertainty. b Left panel: mixed linear model analyses revealed significantly negative associations between d factor score estimates and the mean z-score of FA in WM tracts (β = –0.033, p = 4.08 × 10−7, 97.5% CI, [−0.045, −0.020]) and between d factor score estimates and the mean z-score of GM (β = −0.036, p = 1.80 × 10−5, 97.5% CI, [−0.052, −0.020]). Middle panel: the d factor score estimate showed significantly positive associations with the number of extreme negative deviations of FA (β = 0.008, p = 1.75 × 10−5, 97.5% CI, [0.004, 0.012]) and GM volume (β = 0.011, p = 2.96 × 10−5, 97.5% CI, [0.006, 0.016]). Right panel: mixed linear model analyses were used to identify the relationships between d factor score estimates and extreme positive deviation of FA and GM volume. There are no significant associations between d factor score estimates and extreme positive deviation of FA (β = 0.003, p = 0.653, 97.5% CI, [−0.011, 0.017]) while there are significantly negative associations between d factor score estimates and extreme positive deviation of GM volume (β = −0.003, p = 0.007, 97.5% CI, [−0.005, −0.001]). The extreme deviation is defined as extreme deviation |z | > 2.6. All tests are two-sided.

Back to article page