Efficient probabilistic representation of tibiofemoral soft tissue constraint

Abstract

Verified and efficient representations of knee ligamentous constraints are essential to forward-dynamic models for prediction of knee mechanics. The objectives of this study were to develop an efficient probabilistic representation of knee ligamentous constraint using the advanced mean value (AMV) probabilistic approach, and to compare the AMV representation with the gold standard Monte Carlo (MC) approach. Specifically, the effects of inherent uncertainty in ligament stiffness, reference strain and attachment site locations on joint constraint were assessed. An explicit finite element model of the knee was evaluated under a series of anterior–posterior (AP) and internal–external (IE) loading at full extension and 90° flexion. Distributions of AP and IE laxity were predicted using experimentally-based levels of ligament parameter variability. Importance factors identified the critical properties affecting the predicted bounds, and agreed with reported ligament recruitment. The AMV method agreed closely with MC results with a four-fold reduction in computation time.