Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Industrial and Manufacturing Systems Engineering

First Advisor

Gary A. Mirka


EMG-assisted biomechanical modeling is a well-established modeling technique for estimating muscle forces in biomechanical models of the lumbar spine. Fatigue, however, creates a problem in that fatigue alters the relationship between the EMG signal generated by the muscle and the amount of force being generated. The aim of this study was to evaluate the impact of fatigue on EMG-related components in an EMG-assisted biomechanical model: the gain factor (i.e. maximum muscle stress value in N/cm2), force-length modulation factor, and force-velocity modulation factor. This is a particularly relevant research topic in that fatigue is considered a potential risk factor for musculoskeletal disorders, and being able to quantify muscle forces and joint reaction forces in these fatigued conditions would be helpful to understand the underlying risk factors in these types of exertion. The present study can inform and guide efforts in determining safety criteria in task design to decrease incidences of musculoskeletal disorders. This study was conducted in two phases: the isometric extension phase (1) and the isokinetic extension phase (2). Each was designed to provide the data necessary to evaluate the hypothesis that either the length-force modulation factor (Phase 1) or force-velocity modulation factor (Phase 2) need to be dependent on the level of fatigue experienced by the extensor muscle of the lumbar region.

Four subjects participated in each of these phases, performing trunk extension exertions at a level of 50% of their maximum force generation capacity to generate muscular fatigue in the trunk extensor muscles. In the isometric phase subjects performed controlled, isometric trunk extension test contraction exertions at 10, 20, and 30 angles of trunk flexion on three different days. In the isokinetic phase, subjects performed controlled, isokinetic trunk extension test contraction exertions at 5 and 15 degrees/sec angular velocity (concentric range of motion) on two separate days. As they performed these exertions, EMG data were collected from the longissimus, iliocostalis, multifidus, latissimus dorsi, external obliques, internal obliques, and rectus abdominis muscles.

The EMG results of the trunk extensors (the fatigued muscles) obtained from the isometric and isokinetic phases of this study indicate that the force-length and force-velocity modulation factors are independent of fatigue for the levels of trunk flexion angle and trunk extension velocity considered. The results showed that, while there were strong effects of fatigue, trunk flexion angle, and trunk extension velocity as main effects, there was no significant interaction between trunk angle and fatigue in the Phase 1 study and there was no significant interaction between the trunk extension velocity and fatigue in the Phase 2 study. Significant interactions of these variables would imply that the fatigue was differentially affecting these responses. These non-significant results imply that there is no need to adjust the force-length and force-velocity modulation factors in an EMG-assisted biomechanical model in the range of the levels of the independent variables tested.

The third hypothesis considered in this experiment focused on the need to have a fatigue-dependent gain factor in an EMG-assisted biomechanical model. It is well known that the onset of fatigue leads to a decline in the force generation capability in fatigued muscle, and it is therefore a reasonable hypothesis that the gain factor of this muscle would differ as a function of fatigue level. Using regression analysis, the ratio of pre-fatigue gain factor and fatigued gain factor was shown to be correlated well with the ratio of the pre-fatigue and fatigued median frequency values of the primary trunk extensor muscles. This led to formal modification of the gain of the extensor muscles based on the apparent decrease in median frequency and its pre-fatigue gain factor. The results of this analysis revealed that mean normalized error between predicted and measured internal moments improved from 17.5% error to 9.6% error via the implementation of this modified, fatigue-dependent gain factor. When used to calculate spine reaction forces, the modified EMG-assisted model output shows that the compression force on L5/S1 doesn't change during fatigue development while anterior-posterior shear force increases.

To provide some validation of this new model, two new subjects were recruited to perform different fatiguing lifting exertions. These subjects performed lifting tasks for 20 and 25 trunk flexion angles on two different days in the isometric extension phase, as well as 5 and 10 degrees/sec angular velocity (concentric range of motion) in the isokinetic extension phase on two separate days. Mean normalized error between predicted and measured internal moments affirms that model accuracy improved significantly from 21.4 to 12.9%. Again, applying these results in the calculation of the spine reaction forces, the model with invariant gain factor leads to increases in estimated compression force as fatigue develops, while the modified model shows that the compression forces on L5/S1 are effectively constant. Use of this newly-developed EMG-assisted model could lead to more accurate estimates of spine loading in a manner that requires only gain factor modification without alteration of force-length and force-velocity modulation factors.

The results are particularly important for ergonomists interested in understanding spinal loading and injury risk under fatiguing conditions. This improved EMG-assisted biomechanical modeling technique may help in the establishment of better safety criteria for occupations that generate significant low back muscle fatigue.

Copyright Owner

Omid Haddad



Date Available


File Format


File Size

164 pages