Computational models of biomechanical systems can provide valuable insight into joint functionality and implant performance under conditions that are difficult to test in vivo or using in vitro experimental setups. The main objective of this research was to develop and demonstrate the feasibility of explicit finite element (FE) based models to address current concerns in biomechanics. Specifically, efficient probabilistic methods were applied to evaluate the effects of inherent system uncertainties on output measures, and custom user subroutines and optimization algorithms were incorporated within the FE framework to develop muscle-actuated dynamic models.

A probabilistic FE model of the Stanmore knee simulator was developed. Inputs to the model included International Organization for Standardization (ISO) force-controlled gait loading conditions, along with normal distributions of component alignment, loading and environmental parameters representing uncertainty associated with the experimental setup. The model predicted distributions of output measures, including 1 and 99 percentile bounds of joint kinematics, range of motion, peak contact pressure, and linear and gravimetric wear. Sensitivity of the performance metrics to input parameters were reported to identify the important input parameters.

Three-dimensional musculoskeletal models of the lower extremity were developed. A probabilistic model was used to evaluate the effects of muscle attachment location uncertainty and joint kinematic variability on moment arm predictions from different estimating techniques. Muscle-actuated dynamic FE models were created to simulate activities driven by direct forces or forces calculated from muscle parameters and activations data. Inputs to the FE models included patient anthropometry data, muscle and soft-tissue properties, contact definitions, and muscle parameters based on a Hill-type model. The models were optimized for muscle forces or activations to reproduce kinematic signatures representing knee flexion and chair rise activities. Outputs of the models included joint reaction forces, ligament loads, and tibiofemoral and patellofemoral kinematics and contact pressures. The computational framework developed represents a novel methodology to perform comparative studies of joint behavior and implant designs.