Tai Chi Practice Promotes Network Flexibility And Mitigates Age-Related Decreases In Dynamic Functional Connectivity

Although Tai Chi (TC) has previously been shown to offer compensatory responses to normal age-related functional connectivity (FC) decrements, such responses remain poorly characterized. PURPOSE: To compare and characterize the magnitude of aging-related versus TC-related differences using dynamic FC. METHODS: High-density EEG data (64 channels) were collected from TC older adult practitioners (TCOA, n = 15, age: 65-75 years), age-matched controls (OAC, n = 15), and young adult controls (YAC, n = 15, age: 18-30 years) during resting state conditions. Source-localized EEG data were fit to a hidden Markov model to extract recurrent neural network dynamics. Dynamic functional connectivity (FC) was calculated for intra- and inter-network values and categorized by FC changes. Network-based statistics identified component-level group differences. Wilcoxon Signed Rank Tests compared aging effects (OAC vs. YAC) to TC-related effects (OAC vs. TCOA). RESULTS: For intra-network connectivity, TC-related differences exceeded aging differences in decreased magnitude of negative correlations (W = 459, p = 0.036, d = -0.32), decreased magnitude of positive correlations (W = 425, p = 0.026, d = -0.32), and transitions from positive to negative correlations (W = 3058, p = 1.03E-6, d = -0.43). For inter-network connectivity, TC-related differences were larger in decreased magnitude of negative correlations (W = 13398, p < 2.80E-10, d = -0.40), decreased magnitude of positive correlations (W = 14407, p = 8.33E-10, d = -0.38), transitions from negative to positive correlations (W = 68605, p = 1.28E-14, d = -0.28) and transitions from positive to negative correlations (W = 51667, p = 1.28E-14, d = -0.33). These patterns suggest that TC practice induces stronger network reorganization effects compared to age-related changes, particularly in inter-network integration. CONCLUSIONS: Our findings reveal distinct patterns of dynamic network reorganization, with TC-related differences mitigating commonly observed patterns of diffuse increases in positive correlations and decreased magnitude of negative correlations while simultaneously promoting greater network flexibility. Future studies should examine network-specific interactions to better understand these neural mechanisms. Supported by: This research was funded, in part, by the Jump ARCHES endowment through the Health Care Engineering Systems Center. Artifical Intelligence was used for code generation, code review, and debugging.