A partially personalized musculoskeletal spine model using MRI and EMG data to compare biomechanics of asymptomatic and low back pain individuals

Altered neuromuscular control and movement strategies influence lumbar spine loading in individuals with chronic low back pain (LBP); however, evidence from validated musculoskeletal models with subject-specific parameterization remains limited. This study estimated trunk muscle forces and lumbar spinal loads in asymptomatic and non-specific chronic LBP individuals using partially personalized models and examined group differences. Static sagittal-plane load-holding tasks spanning low- to high-demand conditions were analyzed in asymptomatic participants and LBP subgroups stratified by pain intensity and disability. Subject-specific muscle cross-sectional areas, segmental masses, upper-limb and external load centres of mass, surface electromyography, and kinematics were directly personalized, whereas spinal geometry, muscle attachment locations, segmental centres of mass, and passive properties were obtained from the reference model using established scaling and regression schemes. Notably, an EMG-assisted framework was further introduced alongside the conventional optimization approach. Model predictions showed strong agreement with in vivo intradiscal pressure. Minimal between-group differences in spinal loads were observed during upright tasks (mostly p > 0.05), whereas the largest differences occurred during high-demand near-full trunk flexion, with LBP groups showing increased shear loading (75-160%, p ≤ 0.07), mainly due to altered lumbo-pelvic kinematics. Sensitivity analyses in upright load-holding tasks showed that removing task-related lumbo-pelvic flexion/extension adjustments increased lumbar loads by 25-33% at higher load positions (hip to head level). Separately, the lack of anatomical personalization resulted in 11-37% higher EMG-assisted estimates across tasks. In conclusion, within the present dataset, altered movement strategies appear to be a key contributor to task-specific increases in lumbar loading, whereas muscle morphology and activation alone showed limited explanatory power.