Figure 1.
The schematic structure of the multi-DOF EMG-driven model.
It comprises of five components: a) Musculotendon Kinematics, b) Musculotendon Activation, c) Musculotendon Dynamics, d) Moment Computation, and e) Model Calibration Process. The multi-DOF EMG-driven model is initially calibrated using the Model Calibration component. After calibration the EMG-driven model is operated in open-loop. Musculotendon units force and the resulting moments are determined as a function of EMG signals and three-dimensional joint angles, without tracking experimental joint moments. Joint moments are predicted with respect to four degrees of freedom (DOFs): hip adduction-abduction (HipAA), hip flexion-extension (HipFE), knee flexion-extension (KneeFE), and ankle plantar-dorsi flexion (AnkleFE).
Figure 2.
Allocation of experimental EMG signals to individual musculotendon units.
The first level of each tree represents innervation zones in the human lower extremity. The second level represents the 18 muscle groups that are innervated from the corresponding innervation zone. At the second level, italic-style written names, connected by a dotted line to the first level, represent muscle groups for which experimental electromyography (EMG) signals could not be recorded (a, b, and g). The remaining 16 groups at the second level represent the muscles from which EMG signals were experimentally recorded. The third level represents the associated musculotendon units (MTUs) within each muscle group. All MTU names were abbreviated from the associated muscle group names in the second level with exception of the biceps femoris long and short head (bicfemlh, bicfemsh), semimembranosus (semimem), semitendinosus (semiten), adductor magnus, longus, and brevis (addmag, addlong, addbrev), and peroneus longus, brevis and tertius (perlong, perbrev, pertert). The gluteus minimus, medius, maximus, and the adductor magnus are modeled by three individual MTU compartments. Within each tree, branches have different colors referring to EMG signals recorded separately.
Figure 3.
First test results: comparing MTU force estimates across single-DOF models.
Square of the Pearson product moment correlation coefficient () and normalized root mean squared deviation (NRMSD) between musculotendon unit (MTU) forces predicted by different single degree of freedom (DOF) models including: hip adduction-abduction (HipAA), hip flexion-extension (HipFE), knee flexion-extension (KneeFE), and ankle plantar-dorsi flexion (AnkleFE) single-DOF models. The MTU names abbreviations are defined in Figure 2. Values for the addmag, gmin, gmax, and gmed have been reported as the average between the individual values associated to the three units each muscle is composed of (Figs. 2a, d, and e).
Figure 4.
Second test results: comparing joint moment estimates between multi-DOF and single-DOF models.
Ensemble average curves with associated standard deviation for the experimental joint moments about four degrees of freedom (DOF) including: hip adduction-abduction (HipAA), hip flexion-extension (HipFE), knee flexion-extension (KneeFE), and ankle plantar-dorsi flexion (AnkleFE). The reported data are from the stance phase with 0% being heel-strike and 100% toe-off events. The ensemble average curves are also reported for the matching joint moment predicted by the four corresponding single-DOF models and by the multi-DOF model. The percentage mean absolute error (%MAE) is reported in a histogram form and quantifies the percentage error between the experimental joint moments and those predicted by the multi-DOF model and by the four single-DOF models respectively. Ensemble average curves and %MAEs are shown for four motor tasks including: walking (WK), running (RN), side-stepping (SS), and cross-over (CO) cutting maneuvers.
Figure 5.
Third test results: Comparing MTU force estimates between single-DOF and multi-DOF models.
Square of the Pearson product moment correlation coefficient () and normalized root mean squared deviation (NRMSD) between musculotendon unit (MTU) forces predicted by the multi-DOF model and by the single-DOF models custom-made to the four degrees of freedom (DOFs): hip adduction-abduction (HipAA), hip flexion-extension (HipFE), knee flexion-extension (KneeFE), and ankle flexion-extension (AnkleFE). MTUs names are defined as in Figure 2. Values for the addmag, gmin, gmax, and gmed have been reported as the average between the individual values associated to the three units each muscle is composed of (Figs. 2a, d, and e).
Figure 6.
MTU force estimates between single-DOF and multi-DOF models.
Ensemble average curves for the force predicted for the tensor fascia latae (tfl) and the rectus femoris (recfem) musculotendon units (MTUs) over 10 running trials. The reported data are from the stance phase with 0% being heel-strike and 100% toe-off events. MTU forces were predicted by the multi-DOF model and by the single-DOF models custom-made to the three degrees of freedom (DOFs) spanned by the two MTUs: hip adduction-abduction (HipAA), hip flexion-extension (HipFE), and knee flexion-extension (KneeFE).