Age‐specificity and generalization of behavior‐associated structural and functional networks and their relevance to behavioral domains

Abstract

Behavior‐associated structural connectivity (SC) and resting‐state functional connectivity (rsFC) networks undergo various changes in aging. To study these changes, we proposed a continuous dimension where at one end networks generalize well across age groups in terms of behavioral predictions (age‐general) and at the other end, they predict behaviors well in a specific age group but fare poorly in another age group (age‐specific). We examined how age generalizability/specificity of multimodal behavioral associated brain networks varies across behavioral domains and imaging modalities. Prediction models consisting of SC and/or rsFC networks were trained to predict a diverse range of 75 behavioral outcomes in a young adult sample (N = 92). These models were then used to predict behavioral outcomes in unseen young (N = 60) and old (N = 60) subjects. As expected, behavioral prediction models derived from the young age group, produced more accurate predictions in the unseen young than old subjects. These behavioral predictions also differed significantly across behavioral domains, but not imaging modalities. Networks associated with cognitive functions, except for a few mostly relating to semantic knowledge, fell toward the age‐specific end of the spectrum (i.e., poor young‐to‐old generalizability). These findings suggest behavior‐associated brain networks are malleable to different degrees in aging; such malleability is partly determined by the nature of the behavior.

Asymmetric generalizability of multimodal brain‐behavior associations across age‐groups

“…On the whole, it appears that our prediction and generalizabilityrelated results were not significantly different across modalities; this is largely consistent with our previous work (Yu & Fischer, 2022).…”

“…On the whole, it appears that our prediction and generalizability‐related results were not significantly different across modalities; this is largely consistent with our previous work (Yu & Fischer, 2022 ). Taken together, these results possibly suggest that different modalities of input features were highly correlated with each other, and hence did not significantly outperform each other in predicting behavioral outcomes.…”

“…Although the choice of atlas can influence accuracy of behavioral predictions, in the current study we were more interested in ratio of the prediction metrics obtained from different age-groups, rather than the absolute prediction metrics in a specific age-group. In this regard, we would expect the use of different atlases to improve or worsen predictions in different age-groups to a similar extent, thus keeping this ratio relatively constant, as demonstrated in a previous study (Yu & Fischer, 2022).…”

“…This suggests that, overall, BBAs derived from the old were significantly more age-generalizable compared to that of the young. Furthermore, one would also notice an obvious disparity between the young-to-middle and young-to-old RMSE ratios (t Previous research has suggested the age-generalizability of brainbehavior associations was partly influenced by the between-agegroup differences in behavioral scores (Yu & Fischer, 2022); that is if the behavioral scores differed tremendously between the young and old, one would expect the young-to-old and old-to-young generalizability to be low. Indeed, our results in Table 2 showed that the RMSE ratios correlated highly with the between-age-group differences in behavioral scores as measured using Cohen's D for most scenarios, except in that of young vs middle age-groups (see Figure 5c,d).…”

“…As expected, this preliminary finding suggested that BBAs are age-specific, they did not generalize well beyond the age-group they were derived from. A more recent study (Yu & Fischer, 2022) modeled the associations between brain connectomes-both resting-state functional and structural connectomes, and a diverse spectrum of behavioral outcomes in a young sample, and showed that when these BBAs are applied to an unseen young and old-age samples, a similar age-specificity effect was observed-behavioral outcomes were predicted more accurately in the unseen young than old age-groups. Furthermore, among the unseen old subjects, cognitive functions were predicted poorly as compared to behavioral outcomes in other domains such as personality and affect.…”