目的  利用图神经网络（GNN）技术对人体肌筋膜系统的复杂连接模式进行模型构建与分析，探讨其在临床诊疗中的潜在应用价值。 方法  招募52名不同年龄段、体型的健康志愿者以及30名慢性腰痛患者（病程>3个月）作为临床验证对象，采集其肌筋膜系统的解剖学与生物力学数据。数据来源包括高分辨率磁共振成像（MRI）、超声弹性成像及生物力学测试，基于这些数据构建全身肌筋膜网络拓扑结构。采用改进的图注意力网络（GAT）算法建立肌筋膜系统的计算模型，对连接强度、应力传递路径及病理状态下网络变化进行定量分析。 结果  数据分析表明，肌筋膜系统具有显著的小世界网络特性，并存在关键连接枢纽。特定的肌筋膜链条在人体运动与力传递中起关键作用。肌筋膜连接模式的异常与慢性腰痛等临床病症密切相关。经临床数据验证该模型预测治疗响应准确率为85.7%。 结论  基于图神经网络的肌筋膜系统模型构建方法为理解人体整体性功能提供了新的视角，为构建基于肌筋膜网络特征的个体化诊疗策略提供了理论基础。

Objective  To model and analyze the complex connectivity of the human myofascial system using Graph Neural Networks (GNNs) and explore their clinical potential.  Methods  Fifty-two healthy volunteers of varying ages and body types, along with 30 patients with chronic low back pain (>3 months), were recruited. Anatomical and biomechanical data of the myofascial system were obtained from high-resolution MRI, ultrasound elastography, and biomechanical testing to construct a whole-body myofascial network. An improved Graph Attention Network (GAT) model was applied to quantify connection strength, stress transmission, and pathological network alterations.  Results  The myofascial system showed distinct small-world network properties with key hub connections. Certain myofascial chains played crucial roles in movement and force transfer. Abnormal connectivity patterns were closely linked to chronic low back pain and cervicobrachial syndrome. The model achieved 85.7% accuracy in predicting therapeutic response.   Conclusion  GNN-based modeling provides a new perspective for understanding human integrative function and supports the development of personalized diagnosis and treatment strategies based on myofascial network features.