An optimization-based biomechanical model of the thoracic spine



Journal Title

Journal ISSN

Volume Title


Texas Tech University


The major objective of this dissertation was to examine formulations of optimization-based models of the thoracic spine, comparing model predictions to EMG data collected during lifting tasks. The models employed here were developed based on traditional lumbar spine modeling techniques, but were expanded to include a representation of the rib cage and to predict forces at multiple vertebral levels.

Optical motion tracking data were used in conjunction with known forces at the hands to calculate reaction moments at vertebrae T8 through T12. These moments were used to generate muscle force predictions using linear and nonlinear models, with and without rib cage representation included, and with and without limitations preventing muscle forces from varying too widely between adjacent vertebral levels. Tasks performed for data collection consisted of symmetric and asymmetric lifts of low and high force loads.

Model predictions were compared to EMG data in order to examine model and test parameters. Though none of the model formulations provided good agreement between model predictions and EMG data, differences in model predictions allowed for comparisons in order to select parameters producing the best results. During this research, the simplest model formulations actually provided the best results. The linear objective function performed better, as did model formulations not including rib cage representation and model formulations not including limitations between vertebral levels. Testing parameters impacted model agreement with EMG as well, with models performing better for male subjects. Better model performance was also found for symmetric lifts and lifts of lower force loads. 



Optimization model, Muscle force prediction, Electromyography (EMG)