Activity Patterns in Relation to Dynamic Functional Network States: A Longitudinal Feasibility Study of Brain–Behavior Associations in Young Adults

Figures

• 

  Figure 1. A subset of sample dynamic double functional independent primitives (ddFIPs), including (a) ddFIP 3, (b) ddFIP 7, and (c) ddFIP 8, along with their associated connectograms, demonstrating the variability and shared structure of dynamic functional interaction patterns across canonical brain networks. Warmer colors indicate stronger positive coupling, whereas cooler colors indicate weaker or negative coupling. Rows and columns correspond to predefined brain networks: subcortical (SC), auditory (AUD), sensorimotor (SM), visual (VS), cognitive control (CC), default mode (DM), and cerebellar (CB).

• 

  Figure 2. State occupancy distributions and representative dynamic connectivity configurations. (a) Pooled histogram of fractional occupancy across all participants and states. The distribution is right-skewed, with the majority of state expressions occurring within the 0–10% range and relatively few instances of sustained dominance. (b) Fractional occupancy (%) for each of the ten ddFIP-derived states across participants. Bars represent mean occupancy, boxplots indicate median and interquartile range, and violin overlays depict the full distribution. State 3 exhibited the highest occupancy and greatest inter-individual variability. (c) Representative connectograms of selected dynamic states (States 3, 6, and 8).

• 

  Figure 3. Associations between dynamic state occupancy and physical activity. Each panel depicts the relationship between occupancy of the ten dynamic connectivity states and physical activity, with individual data points shown across participants and linear fits overlaid. (a) Very active minutes; (b) Lightly active minutes; (c) Sedentary minutes; Across metrics, effect sizes were small, though several trends indicate that higher physical activity and better adherence were associated with greater occupancy of integrative states and reduced occupancy of segregated or sensory-weighted states.

• 

  Figure 4. Associations between dynamic state occupancy and cognitive performance outcomes. Each panel shows linear associations between occupancy of the ten dynamic connectivity states and a cognitive metric, with individual participants represented as points and regression lines overlaid. (a) Age-corrected dimensional change card sort accuracy score at Time 2; (b) List Sorting Working Memory (LSWM) accuracy at Time 2 (c) Flanker inhibitory control and attention accuracy at Time 2.

• 

  Figure 5. Distribution of pairwise Euclidean distances between state amplitudes across all subjects and time windows. The unimodal, right-skewed distribution indicates that most state pairs differ only modestly in amplitude at any given moment, with larger separations occurring less frequently. This empirical distribution was used to guide the selection of convergence and divergence thresholds, with lower-bound thresholds capturing genuinely close similarity and upper-bound thresholds capturing only the most pronounced state separations.

• 

  Figure 6. Convergence-based associations between physical activity measures and dynamic functional connectivity states, quantified using −log10(p)×sign(t) for (a) Very active minutes; (b) Lightly active minutes; (c) Sedentary minutes. Each panel displays the statistical strength and direction of associations between physical activity and pairwise state convergence, reflecting how often pairs of dynamic states exhibit similar amplitudes over time. Higher sedentary behavior was associated with increased convergence among lower-integration states, whereas greater physical activity tended to correspond to reduced convergence between segregated and integrative states, indicating enhanced differentiation of higher-order network dynamics in more active individuals.

• 

  Figure 7. Convergence-based associations between cognitive performance and dynamic functional connectivity states, quantified using −log10(p)×sign(t) for: (a) Age-corrected dimensional change card sort accuracy score at Time 2; (b) List Sorting Working Memory (LSWM) accuracy at Time 2; (c) Flanker inhibitory control and attention accuracy at Time 2. Each panel presents the statistical associations between cognitive outcomes and pairwise state convergence, reflecting the extent to which pairs of dynamic states exhibit similar amplitudes over time. Higher cognitive performance was generally associated with greater convergence among integrative, mid-range states, whereas lower performance tended to correspond to stronger convergence between segregated or sensory-weighted states and executive–default configurations.