Small-world network properties and cortical responses of Tai Chi Yunshou: Insights from fNIRS.

Objective: This study aims to characterize the time-domain features and functional network properties from Tai Chi Yunshou across four states (rest, observation, imagination, and motion) using functional near-infrared spectroscopy (fNIRS), thereby discriminating the neural mechanisms underlying the mental regulation and integrative body-mind harmony of Tai Chi Yunshou.

Methods: 100 healthy participants were included and administered with rest, observation, imagination and motion states of Tai Chi Yunshou. The forty-channel wearable fNIRS system was utilized to evaluate the real-time cortical responses during four states. Brain network properties were analyzed with MATLAB and the GRETNA toolbox. Machine learning models, including random forest, AdaBoost, and gradient boosting, were applied for classification.

Results: Small-world network properties (γ>1 and λ≈1) across four states were detected. Compared to states of rest and observation, the motion state exhibited significantly higher γ, λ, Cp, and Eloc (P < 0.01). Additionally, λ and Cp were significantly greater in the motion state than that of the imagination state (P < 0.01). The motion state had increased betweenness centrality and nodal efficiency hubs of the dorsolateral prefrontal cortex and frontopolar area. Gradient boosting outperformed both AdaBoost and random forest across all evaluation metric.

Conclusion: The motion of Tai Chi Yunshou enhances regulatory capacity in the dorsolateral prefrontal cortex and frontopolar area, boosts local brain processing, and improves network integration. Machine learning models can effectively capture and differentiate the neural features between mental regulation and integrative body-mind harmony.