Coupling characteristics of endocrine regulation and system function under massage interventions

Massage may influence stress related endocrine activity and systemic physiological function, but its dynamic regulatory effects remain difficult to quantify.

This study develops a VariationalMotionBayesian framework to model endocrine related physiological responses under massage intervention. The framework combines mathematical formulation of coupled endocrine and systemic variables, variational latent representation learning, motion field estimation, Bayesian evidence aggregation, and physiological constraint regularization. Multimodal wearable physiological time series are used to infer latent regulatory states and predict individual stress status.

Experiments on WESAD and SWELL-KW show that the proposed model outperforms conventional machine learning, deep learning, and time series classification baselines in Accuracy, Macro F1, Balanced Accuracy, and AUROC. A controlled human massage intervention pilot study further shows coordinated changes in salivary cortisol, salivary alpha amylase, heart rate, RMSSD, electrodermal activity, respiration rate, and skin temperature after massage, and these measured responses are correlated with model derived latent dynamics. These results indicate that the proposed framework can capture physiologically meaningful temporal responses and provide a computational tool for analyzing endocrine related system regulation under massage intervention.