Exploring the impact of spinal kinematics assumptions during daily tasks in musculoskeletal modelling

Background:The precise in vivo motion of vertebrae and the way spinal segments contribute to overall trunk kinematics remain largely uncertain. This lack of direct measurements has led to a variety of simplifications and assumptions within musculoskeletal modelling, where the spine is often represented as a single rigid segment or with a fixed partition of motion across levels. Such modelling choices can significantly influence the predicted joint reaction forces and muscle activations, especially during daily activities involving complex trunk and pelvic motion, such as walking and sit-to-stand transitions. Investigating how different assumptions about spinal kinematics affect the mechanical outcome of the model is therefore essential to improve the physiological realism of computational simulations. Aim: The aim of this work is to explore the impact of different spinal kinematic assumptions on the lumbar, hip and knee joint reaction forces and muscle responses during representative daily tasks using musculoskeletal modelling. Methods: Full-body musculoskeletal models were reconstructed based on existing datasets, recovering the kinematics of two daily tasks: walking (WALK) and sit-to-stand (STAND). On these models, different assumptions of spinal kinematics were implemented in order to explore the influence of vertebral motion modelling on biomechanical outcomes. Five progressive model versions (M0–M4) were developed, each introducing different levels of freedom along the spine: M0- only L5–S1 mobile with all superior intervertebral joints locked; M1- all lumbar (L1–L5) joints mobile with equal redistribution of the lumbosacral motion; M2- lumbar + T12–L1 mobile with redistribution using literature-based coefficients (Bruno); M3- full spine mobile (T1–L5) with literature coefficients (Bruno); M4- full spine mobile (T1–L5) with refined coefficients (“Anatomy standard”). Each model’s kinematics was modified using MATLAB scripts to redistribute joint motion accordingly. Subsequently, the models were processed in OpenSim following a standard analysis workflow, including Scaling, Inverse Kinematics (IK), Inverse Dynamics (ID),and Joint Reaction Analysis (JRA),in order to evaluate the resulting spinal loads. Results: Joint reaction forces varied with spinal kinematics. Versus baseline (M0, only L5–S1 mobile), enabling lumbar–thoracic motion with offset correction (M4) reduced lumbosacral loads: at L5–S1 peak compression fell by ~45% in WALK and ~39% in STAND, while peak shear dropped by 85–91%. Cranial redistribution produced modest increases at upper levels (e.g., shear at L3–L4 in WALK and at L1–L2 in STAND), with most other levels still decreasing. M3 and M4 behaved almost identically (<0.2% difference), indicating that freeing the spine drives the effect more than coefficient refinements. At the lower limb, during WALK hip compression fell ~20% at peak and hip shear ~12% at peak; knee changes were smaller: compression ~-9% peak, shear ~-11% peak. During STAND, hip compression was essentially unchanged (~-2% peak) while hip shear increased (~+6% peak); knee compression and shear were nearly unchanged (|&#916;|&#8818;3%). In summary, introducing spinal degrees of freedom produced a more realistic global alignment and redistributed mechanical loads both along the spine and toward the upper body, highlighting the importance of including spinal motion in full-body musculoskeletal modelling.