The prevalence of low back pain (LBP) in certain subpopulations who experience more extreme repetitive spine kinematics is often attributed to biomechanical factors. Repetitive loading of the spine during lumbar movements produces damage in innervated elements known to be sources of pain, with examples including:​ vertebral body fractures, intervertebral disc tears, endplate lesions, and abnormal disc stress. Thus, understanding how lumbar spine kinematics influence load transfer within the tissues during daily activities can provide insight into determining how population-specific biomechanical LBP develops. Methods used to directly measure lumbar spinal loads in vivo and in vitro can be highly invasive, are not always measured in living humans, and are often not measured during daily activities. Computational modeling techniques provide a solution in silico for estimating lumbar spinal loads that are difficult or impossible to measure. While common types of biomechanical computational models (i.e., musculoskeletal and finite element (FE) models) have many benefits, they are limited when employed separately. Multiscale modeling involves combining two or more computational models into a single framework to leverage their capabilities and enable estimation of tissue loads driven by physiological kinematics and loading.

This research sought to address the need for a computational model of the lumbar spine to estimate tissue-level load transfer driven by in vivo kinematics for people who are prone to developing biomechanical LBP. In this body of work, a multiscale model of the lumbar spine (musculoskeletal + FE) was developed, validated, and applied clinically. The model was validated against in vitro torque-rotation and force-displacement responses, as well as in vivo intradiscal pressure. The model was applied to people with and without a transtibial amputation (TTA) to assess differences in tissue load distribution during daily activities between groups. Lumbar spine tissue loads were also analyzed for a participant with a TTA model can help to corroborate clinical decision making. A design of experiments was performed to determine normal variation in the model and its relation to clinical relevance, as well as technical modeling implications for predicting tissue loads from physiological kinematics and kinetics. The overarching goal of this research was to lay the groundwork for developing a computational model that can be adapted for use with different populations, pathologies, geometries, and activities and can be used in parallel with treatment protocols towards improved patient-specific care of biomechanical LBP.