An innovative neuro-fuzzv modfu for predicting creep ofthf medial collateral ligament

It is well established that progressive recruitment of the collagen fibres in the knee medial collateral ligament (MCL) leads to the non-linear toe-region of the ligament stress-strain curve. It has also been argued that fibre recruitment helps the ligament to lessen and resist creep. Minimal creep in ligaments allows maintaining joint equilibrium. This is especially important for the knee stability in regular daily activities like walking or running where loading is repetitively applied to the joint over many cycles. Nevertheless, due to dependency of fibre recruitment on many factors affecting its behaviour, the level of recruitment of the collagen fibres is difficult to quantify using classical modeling techniques. We therefore developed a soft-computing algorithm to model creep of the knee MCL in two steps: first, the ill-defined fibre recruitment is quantified using the fuzzy systems. Second, the fibre recruitment is incorporated along with creep stress and creep time to model creep using a hybrid neuro-fuzzy system. The model is trained and tested using experimental database including creep tests and crimp image analysis. The model showed very promising results and confirmed the role of fibre recruitment in viscoelastic behaviour of the ligament.