Calibration of neuromusculoskeletal models using functional tasks is performed to calculate subject-specific musculotendon parameters, as well as coefficients describing the shape of muscle excitation and activation functions. The objective of the present study was to employ a neuromusculoskeletal model of the shoulder driven entirely from muscle electromyography (EMG) to quantify the influence of different model calibration strategies on muscle and joint force predictions. Three healthy adults performed dynamic shoulder abduction and flexion, followed by calibration tasks that included reaching, head touching as well as active and passive abduction, flexion and axial rotation, and submaximal isometric abduction, flexion and axial rotation contractions. EMG data were simultaneously measured from 16 shoulder muscles using surface and intramuscular electrodes, and joint motion evaluated using video motion analysis. Muscle and joint forces were calculated using subject-specific EMG-driven neuromusculoskeletal models that were uncalibrated and calibrated using (i) all calibration tasks (ii) sagittal plane calibration tasks, and (iii) scapular plane calibration tasks. Joint forces were compared to published instrumented implant data. Calibrating models across all tasks resulted in glenohumeral joint force magnitudes that were more similar to instrumented implant data than those derived from any other model calibration strategy. Muscles that generated greater torque were more sensitive to calibration than those that contributed less. This study demonstrates that extensive model calibration over a broad range of contrasting tasks produces the most accurate and physiologically relevant musculotendon and EMG-to-activation parameters. This study will assist in development and deployment of subject-specific neuromusculoskeletal models.

使用功能任务校准神经肌肉骨骼模型，以计算特定主题的肌肉肌腱参数，以及描述肌肉激发和激活函数形状的系数。本研究的目的是采用完全由肌肉肌电图 (EMG) 驱动的肩部神经肌肉骨骼模型来量化不同模型校准策略对肌肉和关节力预测的影响。三名健康成年人进行动态肩外展和屈曲，随后进行校准任务，包括伸手、头部接触以及主动和被动外展、屈曲和轴向旋转，以及次最大等长外展、屈曲和轴向旋转收缩。使用表面和肌内电极同时测量 16 块肩部肌肉的 EMG 数据，并使用视频运动分析评估关节运动。肌肉和关节力是使用特定主题的 EMG 驱动的神经肌肉骨骼模型计算的，这些模型使用 (i) 所有校准任务 (ii) 矢状面校准任务和 (iii) 肩胛平面校准任务进行校准和校准。将关节​​力与已发布的仪器植入数据进行比较。在所有任务中校准模型导致盂肱关节力大小比从任何其他模型校准策略得出的数据更类似于仪器植入数据。产生更大扭矩的肌肉比那些贡献更少的肌肉对校准更敏感。这项研究表明，在广泛的对比任务中进行广泛的模型校准会产生最准确和生理相关的肌腱和 EMG 激活参数。这项研究将有助于开发和部署特定主题的神经肌肉骨骼模型。