Modeling of the native knee with kinematic data derived from experiments using the VIVO™ joint simulator: a feasibility study

Despite advances in total knee arthroplasty, many patients are still unsatisfied with the functional outcome. Multibody simulations enable a more efficient exploration of independent variables compared to experimental studies. However, to what extent numerical models can fully reproduce knee joint kinematics is still unclear. Hence, models must be validated with different test scenarios before being applied to biomechanical questions. In our feasibility study, we analyzed a human knee specimen on a six degree of freedom joint simulator, applying a passive flexion and different laxity tests with sequential states of ligament resection while recording the joint kinematics. Simultaneously, we generated a subject-specific multibody model of the native tibiofemoral joint considering ligaments and contact between articulating cartilage surfaces. Our experimental data on the sequential states of ligament resection aligned well with the literature. The model-based knee joint kinematics during passive flexion showed good agreement with the experiment, with root-mean-square errors of less than 1.61 mm for translations and 2.1° for knee joint rotations. During laxity tests, the experiment measured up to 8 mm of anteroposterior laxity, while the numerical model allowed less than 3 mm. Although the multibody model showed good agreement to the experimental kinematics during passive flexion, the validation showed that ligament parameters used in this feasibility study are too stiff to replicate experimental laxity tests correctly. Hence, more precise subject-specific ligament parameters have to be identified in the future through model optimization.